# Flares

> Flares is an AI-powered competitive intelligence platform for B2B teams. It automatically tracks competitors across pricing, features, positioning, messaging, hiring, customer reviews, blogs, ads and social media, then delivers structured competitive reports, weekly action-oriented digests, and live battlecards for sales teams.

Flares is built for product marketing, product, sales and sales enablement teams, plus founders, C-levels, and VC and PE firms tracking a portfolio. Setup takes under 30 seconds: enter a company URL and Flares builds a structured profile of each competitor, then reports what changed. Plans start at $59/month for 3 competitors and go to $129/month for 10, with a 14-day free trial; larger portfolios are priced custom. Unlike enterprise competitive intelligence platforms, it does not assume a dedicated competitive intelligence analyst. Flares also exposes reports, signals and battlecards over MCP, so they can be queried from Claude, Cursor and other MCP-compatible assistants.

## Product

- [Flares — Competitive Intelligence Platform](https://www.flares.tech): Overview of how Flares automatically tracks competitors, captures early signals, and pushes live battlecards.
- [The Flares Method](https://www.flares.tech/method): Flares' opinionated method for running competitive intelligence as a continuous, decision-first system — the principles, the operating loop (Sense → Interpret → Distribute → Act → Learn), and how to implement it. Inspired by how the Flares product works, applicable with any toolset.
- [Interactive Product Demo](https://www.flares.tech/interactive-product-demo): A guided, hands-on walkthrough of the Flares product experience.
- [Competitive Intelligence via MCP](https://www.flares.tech/competitive-intelligence-mcp): Use Flares competitive intelligence directly inside any AI assistant (Claude, Cursor, and other MCP-compatible tools) to get instant, structured answers about competitors.
- [Pricing](https://www.flares.tech/pricing): Plans and pricing for Flares, including the 14-day free trial.

## Research

- [The State of Competitive Intelligence Tools: A G2 Review Study of Klue, Crayon, Kompyte & Contify](https://www.flares.tech/research/competitive-intelligence-tools-g2-review-study): Flares CI Research, 2026. An analysis of 500 verified G2 reviews of the four leading CI tools, coded into 340 distinct themes. Key finding: the only complaint shared by all four tools at material frequency is "too many irrelevant alerts" — the category's defining unsolved problem. Seven capabilities qualify as universal table stakes. Includes per-tool praise/criticism breakdowns and a downloadable PDF.
- [Competitive Intelligence Research](https://www.flares.tech/research): Index of Flares' original, first-party research on the competitive intelligence category.

## Glossary

- [Competitive Intelligence Glossary](https://www.flares.tech/glossary): A reference of 100+ competitive intelligence terms, each with a clear definition, why it matters, and concrete examples.

## Templates

- [Competitive Intelligence Templates](https://www.flares.tech/templates): Free, ready-to-use CI templates. Each is a blank artifact you can fill in immediately: fully visible on the page, copies to the clipboard in a format that pastes cleanly into Google Sheets or Excel, and exports to CSV, Notion/Markdown or PDF.
- [Sales Battlecard Template (Free Competitive Battlecard)](https://www.flares.tech/templates/sales-battlecard-template): A blank, fill-in-ready sales battlecard: header (competitor, their positioning, their pricing, our win rate, last verified, owner), quick take, why we win, why they win, objection handlers, landmine questions, proof assets, and a "do not say" list that keeps reps factual and non-disparaging. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Includes guidance on what belongs in each field and why battlecards go stale.

- [Competitive Analysis Template (Free Competitor Analysis)](https://www.flares.tech/templates/competitive-analysis-template): A blank, fill-in-ready competitive analysis: scope and the decision it informs, competitor snapshot, the buyer's evaluation criteria, capability comparison, pricing and packaging, go-to-market and messaging, customer sentiment, strengths and weaknesses with evidence, and a decisions block with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Includes per-section guidance and the rule that three to five competitors analysed deeply beat twelve analysed shallowly.

- [Competitor Profile Template (Free Competitor Profiling)](https://www.flares.tech/templates/competitor-profile-template): A blank, fill-in-ready profile of one competitor: header with owner, threat level and verified date; company snapshot; product and capabilities with a maturity column; pricing and packaging; positioning and target customers; go-to-market motion; strengths and weaknesses; customer sentiment with frequency; recent signals and trajectory; and implications with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Includes per-section guidance, the public-sources-and-honest-methods rule, and the distinction between a profile, a competitive analysis and a battlecard.

- [Competitive Intelligence Report Template (Free CI Report)](https://www.flares.tech/templates/competitive-intelligence-report-template): A blank, fill-in-ready periodic CI report: header (period, audience, competitors covered, owner, next report due); an executive summary that names the decision being asked for; what changed this period; product and feature moves; pricing and packaging changes; positioning and messaging shifts; field intelligence from won, lost and stalled deals; market and category signals; a threat assessment with movement since last report; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: report the delta rather than the standing state, label every claim Confirmed/Likely/Unconfirmed, and open the next report with what happened to the last one's decisions. Also answers, honestly, that there is no established "7 Ps of competitive intelligence" framework: the query refers to the 7 Ps marketing mix.

- [Win/Loss Report Template (Free Win-Loss Analysis Report)](https://www.flares.tech/templates/win-loss-report-template): A blank, fill-in-ready win/loss report: header stating how many deals were reviewed and how they were selected; executive summary with win rate and counts; deal-by-deal log holding the buyer's verbatim reason; reason analysis counting each driver in wins as well as losses; head-to-head win rate per named competitor; where deals were decided, by stage, with whose problem it is; buyer verbatims; movement since the last report; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: ask the buyer rather than the rep, review wins as carefully as losses, keep no-decision deals in the analysis but out of the win rate, and treat "we lost on price" as a claim to verify. States plainly that there is no credible universal win-rate benchmark, and disambiguates Excel's win/loss sparkline from win/loss analysis.

- [Competitive Landscape Template (Free Competitive Landscape Analysis)](https://www.flares.tech/templates/competitive-landscape-template): A blank, fill-in-ready competitive landscape: market defined as the job the buyer is hiring for, plus who is deliberately out of scope; a player map spanning all four competitor types (direct, indirect, replacement, potential); segment-by-segment leadership; positioning axes for a 2x2; scale and momentum signals instead of unsourced market share; the five forces shaping the market; whitespace with evidence and an honest reason it is still open; a triggered watch list; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: the market boundary causes more errors than anything inside it, the status quo (a spreadsheet, a manual process, doing nothing) is usually the largest competitor, and never publish a market share figure you cannot source. Includes a specific answer on which parts of Porter's five forces still hold up for software markets and which do not.

- [Competitive Intelligence Executive Brief Template (Free One-Page Brief)](https://www.flares.tech/templates/competitive-intelligence-executive-brief-template): A blank, fill-in-ready one-page executive brief for competitive intelligence: header naming the specific decision-maker and the decision deadline; a bottom-line-up-front (BLUF) block of four sentences; the three things that changed, each with a confidence label; business impact sized in numbers leadership already tracks; two or three options including doing nothing, each with cost, risk and a marked recommendation; the ask, with an owner once approved and what happens if we do nothing; and a pointer to the fuller analysis. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: one page as a hard constraint, conclusion before evidence, size the impact or say you cannot, and never present a problem without options. Distinguishes the executive brief (one page, for leadership, ends in an ask) from the full competitive intelligence report.

- [Objection Handling Template (Free Sales Objection Handling Matrix)](https://www.flares.tech/templates/objection-handling-template): A blank, fill-in-ready objection handling matrix: header naming where reps actually access it; six objection categories with what each sounds like and who resolves it; the core matrix pairing the buyer's verbatim words with the concern underneath, the response and the proof asset; diagnostic questions to ask before responding; competitor-specific claims with an honest true/partly-true/false verdict; price and budget objections with when to hold and when to concede; a do-not-say list; a practice and rollout plan; and a block routing recurring objections to whoever can remove them. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: ask one clarifying question before answering, concede partly-true competitor claims first, never write a response you would not say out loud, and fix recurring objections rather than answering them forever. States plainly that the "4 P's of objection handling" is not a canonical framework (two incompatible versions circulate with no attributable origin), and describes the AAA model as acknowledge/address/ask with acknowledged variants.

- [Feature Gap Analysis Template (Free Competitive Gap Analysis)](https://www.flares.tech/templates/feature-gap-analysis-template): A blank, fill-in-ready feature gap analysis: scope naming the decision and where the gaps came from; the gaps stated in the buyer's words with dated evidence; deal impact separating "appeared in" from "decided"; severity by buyer priority with blocker-versus-preference and the existing workaround; why each gap exists and whether it was a deliberate trade-off; an explicit response per gap (build, partner, counter-position, accept, disqualify) with the sales talk track; where you lead and whether it is defensible; gap movement between reviews; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core position: parity is not a strategy, and a team that closes every gap a competitor opens has handed them the roadmap; most gaps should end in partner, counter-position or deliberate acceptance. Distinguishes feature gap analysis (the decision layer) from a feature comparison matrix (the grid), and disambiguates the SERVQUAL five-gaps service-quality model and the four classic business-planning gap types, neither of which is a feature gap framework.

- [Competitor Tracking Spreadsheet Template (Free Competitor Tracker)](https://www.flares.tech/templates/competitor-tracking-spreadsheet-template): A blank, fill-in-ready competitor tracker built around change over time rather than a static grid: tracker scope and who reads it; the roster tagged by competitor type and tier; what you track and where, one row per source with an honest check frequency; a short current baseline; an append-only dated change log with evidence links; so-what triage with confidence and urgency; distribution matched to audience; a monthly coverage health check that catches broken alerts; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: a tracker is a change-detection instrument, not a snapshot; append to the log rather than overwriting the baseline, because the sequence is the asset; triage before distributing, since irrelevant alert volume is the one complaint shared by all four leading CI tools; and treat prolonged silence from a competitor as a broken alert until proven otherwise. Explains the difference between a competitor tracking spreadsheet and a competitive analysis spreadsheet, and the honest signals that you have outgrown a spreadsheet.

- [Competitive Benchmarking Template (Free Benchmark Analysis)](https://www.flares.tech/templates/competitive-benchmarking-template): A blank, fill-in-ready competitive benchmarking framework: scope naming the benchmarking type and the decision it informs; the benchmark set with what each peer is and is not comparable on; metric definitions written before any data is collected; the benchmark table with best-in-set called out; a comparability and data-quality section rating the confidence of every peer figure; position, gap and direction since the last round; the practice behind each gap; targets with a reason for not defaulting to parity; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: most wrong benchmarks are wrong at the definition stage rather than the data stage; you instrument yourself and infer competitors, so label the inferences; put at least one non-competitor in the set because direct rivals are usually weak where you are weak; and a gap with no identified mechanism cannot be closed by decision. Attributes the four benchmarking types (internal, competitive, functional, generic) to Robert C. Camp's 1989 Xerox work, states plainly that there is no canonical seven-step benchmarking model, and identifies the "McKinsey competitor matrix" as the GE-McKinsey nine-box portfolio tool, which was not built for comparing rivals.

- [Pricing Teardown Template (Free Competitor Pricing Teardown)](https://www.flares.tech/templates/pricing-teardown-template): A blank, fill-in-ready teardown of one competitor's pricing model: scope; published pricing recorded verbatim with screenshots; the value metric the bill actually scales with; packaging and gates mapped to the capabilities that come up in your deals; the real entry price including onboarding, minimums and mandatory add-ons; discounting and negotiation behaviour filled only from deals you were in; an append-only log of price and packaging changes; implications for your own pricing including where their model beats yours; and decisions with owners and dates. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: the value metric matters more than the headline price and can change while the list price stays identical; never state a competitor price without a source and a date; leave the discounting section empty rather than guessing; and most teardowns should change what sales says, not what you charge, since matching a price cut converts a positioning problem into a margin problem. Disambiguates "teardown" from its construction/demolition meaning, and covers the $9 price-ending evidence (Anderson & Simester, Quantitative Marketing and Economics, 2003) along with why catalogue-retail pricing psychology does not transfer to negotiated B2B software.

- [Messaging Framework Template (Free Competitive Messaging)](https://www.flares.tech/templates/messaging-framework-template): A blank, fill-in-ready messaging framework anchored to competitors rather than to brand values: scope; audience and competitive alternatives including the spreadsheet and the do-nothing option; a positioning statement (for whom, in what category, delivering what value, unlike which alternative, credible because of what); three value pillars each tested against "could a competitor say this?"; dated proof per pillar; a messaging matrix by audience where only genuinely different cells are filled; competitive contrast quoting what rivals actually claim; a do-not-say list with replacements; and decisions naming the published assets that change. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: source the language from win/loss interviews and call recordings rather than from an internal workshop; differentiation is meaningless without a named alternative; a pillar with no proof is an adjective; and a framework that changes no published copy has not shipped. Distinguishes positioning from messaging using April Dunford's five components, states that there is no canonical set of five brand pillars or three types of messaging, and disambiguates marketing messaging from messaging APIs and notification templates.

- [Competitor One-Pager Template (Free One-Page Competitor Brief)](https://www.flares.tech/templates/competitor-one-pager-template): A blank, fill-in-ready one-page competitor brief for a rep with three minutes before a call: scope naming the reader, the moment and the one sentence they must remember; a six-fact snapshot including head-to-head win rate with sample size and the two-quarter trend; what they sell with the real entry price and value metric; three genuine strengths with the deals lost to each; three sourced weaknesses each paired with the question that surfaces it; exactly three comparison rows with the line a rep can say; four situational plays with an explicit "do not do this"; and a cut list recording what was left off and where it lives instead. Each section ships one example to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: the one-page constraint is the point, so cut rather than shrink the type; write why they win before why they lose, or reps discard the page; never state a weakness you cannot source, because buyers check; and put the verification date on the visible page. Distinguishes the one-pager (a three-minute briefing) from the battlecard (in-call ammunition) and the competitor profile (the full research file), and explains why it should never be shared outside the company.

- [Competitor Teardown Template (Free Product Teardown)](https://www.flares.tech/templates/competitor-teardown-template): A blank, fill-in-ready teardown of a competitor's product based on first-hand use: scope naming one product and one question; access and method recording what your route let you see and what it did not; the first-run experience timed step by step; a core workflow walkthrough framed by the buyer's job rather than by features; capability observations keeping observed behaviour separate from unverified marketing claims; genuine strengths with whether they are adoptable; where the product breaks down, with thresholds and whether it was reproduced; what the pattern of product decisions reveals about their strategy, ending in a falsifiable prediction; and decisions split between sales enablement and product. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: a teardown is primary research, so observed and inferred never share a column; reproduce a weakness before recording it, because once is an anecdote; a teardown that finds nothing impressive was not done carefully; and use access routes you would be comfortable describing, since a finding whose provenance you cannot explain is one you cannot circulate. Disambiguates "teardown" from demolition, covers product and app teardowns, and distinguishes a teardown (first-hand, point-in-time) from a competitor profile (assembled from published sources).

- [Sales Call Prep Sheet Template (Free Call Planning)](https://www.flares.tech/templates/sales-call-prep-sheet-template): A blank, fill-in-ready prep sheet for ONE sales meeting: call details naming the outcome you want in a sentence and what would make the call a failure; an account snapshot where every fact carries a source and a confidence level; a room map recording what each attendee is personally measured on; deal history including your own unmet commitments; the competitive situation limited to alternatives genuinely in play, including the status quo and doing nothing; five to eight ranked questions each paired with what a good answer sounds like; objections predicted for these specific people with what is really behind each one; a call plan costed in minutes against the time you will actually get, marking what to cut; and next steps drafted before the call with named owners on both sides. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: fifteen minutes of preparation, not an hour; source every fact or move it into your questions; prepare the room rather than the account, since four attendees usually mean four different numbers; and protect the last five minutes for next steps. Distinguishes a prep sheet (one meeting, written by the rep, disposable) from a battlecard (one competitor, maintained centrally, reused). Confirms that the seven steps of selling are genuinely canonical (Dubinsky, 1980-81; revisited by Moncrief and Marshall, Industrial Marketing Management, 2005) while the six-step and eight-step variants are compressions of it, and states plainly that the "5 C's of sales", the "7 golden rules of sales" and the "3-3-3 rule" have no attributable origin. Disambiguates the 30-60-90 plan (rep onboarding, not a call framework) and the ghost call (a VoIP fault, not a sales technique).

- [Competitive Differentiation Matrix Template (Free Download)](https://www.flares.tech/templates/competitive-differentiation-matrix-template): A blank, fill-in-ready matrix that scores weighted buyer decision criteria rather than features: scope fixing ONE segment, since criteria and weights differ between segments; six to nine criteria framed as buyer questions with weights that must total 1.00; a scored matrix rating every alternative 1 to 4 including the status quo, with weight times gap ranking the differences; evidence and confidence behind each score; where you genuinely win, with how fast a competitor could copy it; where you genuinely lose, quantified from loss data; an explicit close, neutralise or concede decision per gap; the sentence a rep says with its proof and the version never to say; and decisions split between positioning and product. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: weights come from buyers and scores come from evidence, since this artifact's failure mode is flattery rather than inaccuracy; score the status quo and expect it to win rows; never reduce the matrix to one total per vendor, because the shape is the output; and most gaps should be neutralised or conceded, since closing every gap is how products match competitors on everything and lead on nothing. Explains the Competitive Profile Matrix lineage (Fred R. David, Strategic Management: weights summing to 1.0, ratings 1-4, weight times rating), corrects the "McKinsey competitor matrix" query (the GE-McKinsey nine-box is a corporate portfolio tool built for General Electric in the early 1970s, not a competitor comparison tool), covers the "kill sheet" and "competitive kill sheet" synonyms, and states plainly that the "4 types of differentiation", the "4 pillars of competitive advantage" and the "5 steps of a competitive analysis" have no canonical source, while horizontal versus vertical differentiation and Porter's generic strategies do.

- [Win-Back Analysis Template (Free Customer Win-Back)](https://www.flares.tech/templates/competitive-win-back-analysis-template): A blank, fill-in-ready analysis of customers who left AFTER buying, usually for a named competitor: scope fixing one period and one written definition of "lost"; accounts lost with where each went and how you know, marking inferences; stated reason and underlying reason in separate columns, since price is the reason most often stated and least often true alone; patterns counted with the revenue behind each; what has actually changed since they left, with evidence a returning customer could verify independently; a scoring table gating approach on whether the reason is fixed, whether their new contract is near renewal, and whether anyone still replies; the approach per account including what you will NOT offer; win-back economics measured on your own numbers; and decisions separating what fixes the cause from what pursues the accounts. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: ask the customer rather than the account team, because CRM close reasons are systematically biased toward causes outside the owner's control; fix the cause before running any campaign, since approaching a churned customer with nothing new burns the relationship you would need at their next renewal; time the approach to their renewal rather than to your quarter; and separate controllable from uncontrollable churn. Distinguishes win-back analysis (post-sale defection) from win/loss analysis (deal stage). Cites Thomas, Blattberg and Fox, "Recapturing Lost Customers", Journal of Marketing Research, 2004, on reacquisition pricing, and states plainly that the widely quoted "20 to 40 percent probability of selling to a lost customer" traces to Marketing Metrics (Farris, Bendle, Pfeifer, Reibstein) via Griffin and Lowenstein's Customer Winback (2001), describes consumer purchasing, and is not a B2B software benchmark.

- [Competitor Switching Cost Analysis Template (Free)](https://www.flares.tech/templates/competitor-switching-cost-analysis-template): A blank, fill-in-ready analysis of what it costs to move between providers, run in BOTH directions in one document: scope naming one competitor and one segment; the cost for their customer to switch to you, in days and money, recording who inside the buyer bears each cost; the cost for your customer to leave; every cost sorted into the three types; what real observed migrations actually took, measured from decision to full use rather than import to import; what to reduce or absorb, ranked by what buyers cite rather than by what is easy to build; a test separating retention you earn from retention you trap; where the costs are a genuinely durable moat; and decisions split by direction. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: run both directions, because the friction blocking your growth is usually the mechanism protecting your retention; measure one real migration before trusting any estimate, since internal reasoning is consistently wrong in the same direction; count relational costs, which decide more renewals than procedural ones; apply the public-defence test to every retention mechanism, so difficult exports and silent automatic renewals land on the fix list; and assume claimed moats are one motivated competitor away from disappearing. Cites Burnham, Frels and Mahajan, "Consumer Switching Costs: A Typology, Antecedents, and Consequences", Journal of the Academy of Marketing Science, 2003, for the procedural/financial/relational taxonomy, and Porter's Competitive Strategy (1980) for switching costs as a source of entry barriers. Disambiguates switching costs from network effects, and from the accounting "4 types of costs" (fixed, variable, direct, indirect).

- [QBR Competitive Update Template (Free QBR Slide)](https://www.flares.tech/templates/qbr-competitive-update-template): A blank, fill-in-ready competitive SECTION for a quarterly business review, not a whole QBR: update details recording the minutes you will actually get; exactly three ranked headline changes, each translated into what it means for that room; a quarter-over-quarter scoreboard using identical metrics every quarter with both columns shown; competitor moves each carrying a source and a date, including moves that did NOT affect you; where competitors appeared in the pipeline as counts tied to revenue, with data-completeness stated; what you said you would do last quarter and whether it moved the number, including the failures; emerging risks with confidence labelled and the signal that would confirm each; one to three asks with costs, deadlines and what happens if you do nothing; and decisions written in the room with owners, dates and a revisit point. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: report the delta rather than the standing state; build for five to eight minutes even when promised fifteen, since competitive sections sit late and get compressed; keep scoreboard metrics fixed, because changing them destroys the comparison and looks like metric selection; source everything, because whatever you state leaves the room without your caveats; say how complete the data is; and close the loop on last quarter including what failed, which is what makes the room treat your asks seriously. Distinguishes a QBR competitive update (scheduled, quarterly, several competitors, fixed scoreboard) from an executive brief (event-triggered, one issue, written when it happens). Disambiguates QBR from ESPN's Total QBR quarterback rating (introduced 2011, 0-100 scale, distinct from the NFL passer rating that maxes at 158.3), from PI Planning in SAFe, and from OKRs.

- [Competitive Product Comparison Template (Free Chart)](https://www.flares.tech/templates/competitive-product-comparison-template): A blank, fill-in-ready product comparison built to a PUBLISHABLE standard, because this is the one competitive artifact the competitor themselves will read: scope deciding up front whether it will be published, which sets the sourcing bar for every claim; exactly what is being compared, naming tier, billing term, currency and the date each price was read; six to ten criteria framed as buyer questions, including at least two you lose on; the comparison itself, written as the competitor's own documentation would describe them, with no binary ticks for capabilities that are present but shallow or gated; a sourcing log where any claim that cannot be logged is deleted from the comparison; the competitor's best case stated unhedged by their strongest advocate; an explicit when-to-choose-each section containing at least one row that sends the buyer to a competitor; pre-publication checks; and decisions. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: primary sources with dates only, never third-party round-ups; match tiers and billing terms, since mismatched comparison is the commonest error and the easiest for a competitor to expose; concede at least two real advantages, because a table you win outright is discounted wholesale; hold your own claims to the same evidential standard as the competitor's; and re-check quarterly, since an accurate comparison becomes a misrepresentation once stale regardless of intent. Disambiguates the data-visualisation sense of "comparison chart" (bar charts, axis truncation, what every graph needs) and the English-grammar sense of "comparison statement" and "comparison sentence" from the product-marketing artifact. Notes that comparative advertising is regulated differently across the EU, UK and US and should be reviewed by counsel before publication.

- [Competitive Intelligence Request Brief Template (Free)](https://www.flares.tech/templates/competitive-intelligence-request-brief-template): A blank, fill-in-ready INTAKE form a stakeholder completes before any research starts, which is the requirements phase of the intelligence cycle: request details; the decision the answer informs, including what would be done differently depending on the answer (if that row cannot be filled, the decision is already made and the request is seeking justification); three to five questions ranked by decision impact rather than curiosity; what the requester already believes, with provenance and confidence, so the team does not re-discover it; how the answer will be used, naming artifact and audience; effort, source and confidentiality constraints with an agreed ceiling; a written definition of done that explicitly permits "we could not determine this"; the CI team's triage verdict (accept, narrow, refine or decline, always with a reason); and a post-delivery loop recording what decision the answer actually produced. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: no decision, no request (awareness work is legitimate but labelled and scheduled against spare capacity); the requester fills it in, never the analyst; separate the knowable (published pricing, packaging, hiring signals, product behaviour) from the unknowable (internal roadmaps, margins, intentions); partial and on time beats complete and late; declining is part of the job, most often because the answer already exists; and keep the request log, which is the most persuasive artifact a CI function can bring to a budget conversation. States that the intelligence cycle has NO canonical phase count (the CIA describes five, the FBI six), that the "7 Ps of competitive intelligence" and "7 basic principles of intelligence" are not real frameworks, and that the "3x5x2 method" has no established definition, while citing the one that IS real and openly published: Frank Watanabe, "Fifteen Axioms for Intelligence Analysts", Studies in Intelligence vol. 40 no. 5, 1997, CIA Center for the Study of Intelligence.

- [Competitive Intelligence Program Charter Template (Free)](https://www.flares.tech/templates/competitive-intelligence-program-charter-template): A blank, fill-in-ready founding document for a CI function: charter details with a named executive sponsor, whose function shapes the remit more than any other choice; three to five objectives written as outcomes somebody outside the team could confirm, never as activities; what is in and, more importantly, OUT of scope, each exclusion naming the team that owns it instead; stakeholders and the decisions each actually makes, since a stakeholder who cannot name a decision is an audience; deliverables with cadence and service levels committed only at sustainable levels; explicit ethical and legal boundaries listing permitted practices as well as prohibited ones, each prohibition carrying an alternative route; sources and tooling with real costs including internal time; measures based on outcomes and usage rather than document volume; and approval decisions including what was deliberately left open. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: one page beats ten; the exclusions are the operative half; promised-and-missed cadence damages a program more than modest-and-met; measure outcomes and usage, never volume, and name the attribution problem openly rather than claiming a clean causal line to revenue; and cite the charter or it is decorative. Answers "is competitive intelligence legal?" from the Economic Espionage Act of 1996 (18 U.S.C. §§ 1831-1839: trade secret theft is criminal, public-domain assembly triggers no liability, so HOW information was obtained matters more than what it contains) and "is competitive intelligence ethical?" from SCIP's code of ethics (disclose identity and organisation before interviews). States plainly that there is no canonical set of "three basic objectives of a competitive intelligence program" — the widely repeated trio traces to homework-answer sites, not to SCIP.

- [Business Wargaming Template (Free Business Wargame Plan)](https://www.flares.tech/templates/business-wargaming-template): A blank, fill-in-ready plan for running a competitive war game, a structured role-play where colleagues play your named competitors against a real plan: wargame details tied to a pending decision with a decider and a date; the one question the game must answer, which must be one analysis genuinely cannot settle; three to five teams including a customer team, with named companies rather than archetypes; ASYMMETRIC intelligence packs giving each team only what that player would plausibly know, with the exclusions written down; two or three tightly timed rounds where each team hands in ranked moves with triggers; written rules read aloud at the start; a live move and counter-move log capturing the REASONING behind each move, not just the move; a debrief run in the room asking what surprised people rather than what they learned; and decisions written before anyone leaves. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: no pending decision, no war game; research anything published sources could answer before role-playing it; put senior sceptics on the COMPETITOR teams rather than the home team, since home teams merely defend; never give competitor teams your roadmap or margins, because responses built on information no real rival would have are fiction dressed as rigour; log reasoning because it becomes a monitoring trigger months later; and route the output into research, since a war game produces plausible hypotheses rather than verified fact. Grounded in Benjamin Gilad's Business War Games (with Todd Stitzer), the standard treatment of the method: roughly one to one and a half days, 12 to 48 participants, teams of three to eight, a host team presenting the plan and competitor teams playing devil's advocate. Disambiguates business wargaming from tabletop and miniature wargaming (the hobby that dominates the bare keyword), from scenario planning (environment rather than opponents), and from red teaming (broader, and often one person).

- [Market Share Analysis Template (Free Calculation)](https://www.flares.tech/templates/market-share-analysis-template): A blank, fill-in-ready share analysis that starts with the DENOMINATOR, because most market share figures are unusable not from bad arithmetic but because the market was never defined: scope stating the market in one sentence a competitor could dispute, the period, and whether it is measured in revenue, units, seats or customers; each market boundary decision recorded with its effect on your own share, including whether unpriced alternatives count; the denominator sized two ways, one bottom-up and one top-down, reporting the gap rather than averaging; share by player with an unattributed remainder row and relative share against the leader; share by SEGMENT, where the real findings are; share movement attributed to something observable or marked unexplained, and labelled one-off or structural; market concentration; a confidence table marking every figure measured or estimated; and decisions. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: define the market first and keep the definition constant, since share only means anything as a trend; include unpriced alternatives, because in many B2B categories spreadsheets and internal builds win more deals than any named vendor; never force shares to sum to 100%; work at segment level, since offsetting movements read as stability in the aggregate; mark measured against estimated on every row; and prefer direction over level when confidence is low, because errors are correlated between periods and largely cancel. Covers relative market share (BCG growth-share matrix: your share divided by the leader's, so 25% against a 50% leader is a fundamentally different position from 25% against a 15% runner-up) and the Herfindahl-Hirschman Index with the 2023 DOJ/FTC merger guideline thresholds. Corrects the "McKinsey market analysis framework" query: the 7-S model (Peters, Waterman and Phillips, "Structure Is Not Organization", Business Horizons, 1980) is about organisational effectiveness, not market sizing.

- [TAM SAM SOM Analysis Template (Free Market Sizing)](https://www.flares.tech/templates/tam-sam-som-template): A blank, fill-in-ready market sizing built BOTTOM-UP with the arithmetic visible: scope naming the audience, since an investor sizing and an internal build decision are held to different standards; each of the three layers defined for your business with its exclusions written down; TAM built by counting the addressable population and multiplying by observed contract value rather than list price, every input sourced; SAM narrowed only by constraints that apply TODAY, each recorded with whether removing it would be a funded project; SOM built from sales capacity and measured win rates and bounded by a stated period; an assumption register so a reader can attack the weakest input rather than the conclusion; a sensitivity check varying the two or three inputs that move the answer most; cross-checks against observable quantities; and decisions. Each table ships one example row to delete. Copy to clipboard (pastes into Sheets/Excel with columns intact) or download CSV, Notion/Markdown or a fillable PDF. Core principles: never derive SOM as a chosen percentage of SAM, since a percentage has no mechanism behind it and cannot be defended or improved; a constraint you plan to remove is still a constraint, so keep aspirational scenarios separate and labelled; treat published category totals with real caution and use them only as a top-down cross-check; run the sensitivity check, which routinely reveals the whole answer rests on one unverified input; cross-check SOM against the market leader's estimated revenue, because most sizing errors are order-of-magnitude; and keep the cross-checks that FAILED in the document, since they are the strongest evidence the work was done carefully. States plainly that TAM SAM SOM is practitioner vocabulary rather than a canonical framework — no single originator, popularised through startup literature including Steve Blank and Bob Dorf's The Startup Owner's Manual — so there is no authority to appeal to on where the boundaries belong. Disambiguates SOM in auditing (summary of misstatements) and PAM (not part of the standard three layers).

## Guides

- [Competitive Intelligence Guides](https://www.flares.tech/guides): In-depth, evergreen guides on running competitive intelligence — tracking competitors, building battlecards, win/loss analysis, choosing tools, and turning signals into decisions.
- [Competitive Intelligence Software: How to Choose the Right Platform (2026)](https://www.flares.tech/guides/competitive-intelligence-software): A buyer's guide to CI software. Most tools clear the same table-stakes bar (automated monitoring, single source of truth, battlecards, reports); the deciding factors are signal relevance, content freshness, and AI/MCP distribution. Sourced by Flares' G2 review study of Klue, Crayon, Kompyte & Contify, whose one universal complaint was "too many irrelevant alerts."
- [Competitive Intelligence with AI: What It Solves, and What It Makes Worse](https://www.flares.tech/guides/competitive-intelligence-with-ai): AI solves collection and synthesis, but collection was never the bottleneck. Flares' study of 500 verified G2 reviews found "too many irrelevant alerts" is the only complaint shared by all four leading CI tools (Klue 3.3%, Crayon 10%, Kompyte 5%, Contify 6%), and reviewers name the automated filtering itself as a weak point. Argues AI's real value in CI is interpretation and distribution, not collection.
- [How to Do a Competitive Analysis with AI (Without Publishing Wrong Facts)](https://www.flares.tech/guides/competitive-analysis-with-ai): A verification-first, six-step workflow for running a competitive analysis with ChatGPT, Claude, Gemini, Perplexity or Mistral. Includes twelve copy-ready prompts, how the assistants differ, what you should never paste into a public AI account, and a verification checklist to keep fabricated competitor pricing and features out of battlecards.
- [Competitive Intelligence MCP: Put Competitor Data Inside Your AI Assistant](https://www.flares.tech/guides/competitive-intelligence-mcp): How the Model Context Protocol brings battlecards, competitor reports and live signals into assistants like Claude and ChatGPT. Covers the real use cases (battlecard before a call, agent-drafted stakeholder summaries, CRM enrichment, pushing signals in from external sources), setup without code, scoping and security, and the honest limits: MCP fixes distribution, not monitoring or relevance.

## Sources — where competitive intelligence comes from

- [Competitive Intelligence Sources](https://www.flares.tech/sources): Index of where competitive intelligence actually comes from. Each page answers one "how do I find X" question by naming every real source, grading its reliability for that specific data point, giving the manual method, and stating honestly what no source will tell you. Every page carries a five-column source comparison table (source, what it gives you, cost, how current, reliability graded High/Medium/Low), an ordered manual procedure, a verification section, a legal-boundary section, and a "what you cannot find and the best proxy" section. All methods are manual: the cluster never recommends scraping or automated extraction.
- [How to Find Competitor Pricing: 9 Sources, Ranked by Reliability](https://www.flares.tech/sources/competitor-pricing): Nine sources for competitor pricing: their pricing page, archived pricing pages via the Internet Archive (price history), cloud marketplace listings and public sector contract awards (both carry real transaction prices and are the way around a "Contact sales" page), your own closed-lost deals (the only source of street price rather than list price), review sites, partner/reseller price lists, app store listings, and paid procurement benchmarks. Verification rests on normalising currency, term, seat minimum and region before comparing, and on separating list price from street price as two fields. Legal section covers the four boundaries: no misrepresenting yourself to get a quote, no confidential pricing, no hiring for pricing knowledge, and never exchanging pricing intentions with a competitor (price fixing, prosecuted in the EU and US). Cannot be found: discount floor, custom enterprise contract value, average selling price, future pricing — each with a stated proxy.
- [How to Find Competitor Revenue: 10 Sources and How Wrong Each One Is](https://www.flares.tech/sources/competitor-revenue): Sorts every revenue source into three tiers — Filed (audited, submitted to a regulator), Stated (public claims by the company) and Estimated (models) — and argues that mixing them unlabelled is the core error in competitive financial analysis. Key point most US-centric advice misses: private companies across most of Europe must file annual accounts that become public, so a filed figure is often available where people assume they must estimate. Includes a jurisdiction table (UK Companies House, EU business registers, Bundesanzeiger, France, SEC EDGAR). Covers the bottom-up build (verified price x estimated customer count) as the only defensible estimate, and states plainly that the widely repeated "employees x $200,000" rule can be out by a factor of two or three and is a plausibility check, never an estimate. Cannot be found: private-company ARR, revenue by product line, churn and net revenue retention, profitability for small filers.
- [How to Find a Competitor's Customers: 10 Sources That Name Names](https://www.flares.tech/sources/competitor-customers): A competitor's customer list is assembled from public artefacts, never found in one place. Ten sources: logo wall and case studies, archived versions of those pages, review sites, job postings naming their product, LinkedIn profiles, technology detection reverse lookup, public contract awards, integration and partner directories, events and communities, and your own CRM and win/loss records. The differentiating technique is the archive diff: comparing their customer page against six- and twelve-month-old snapshots to list logos that have quietly disappeared, which is a churn shortlist rather than proof. Verification requires two independent signals and separates "customer" from "happy customer" as different fields. Legal section covers trade-secret exposure from lists that were taken rather than assembled, GDPR and CAN-SPAM constraints on contacting people, and inducing breach of contract.
- [How to Find a Competitor's Employee Count: 8 Sources, Compared](https://www.flares.tech/sources/competitor-headcount): Eight sources, with the key insight that they measure different populations and will not agree. Filed accounts are authoritative: SEC filings state employee count under Regulation S-K human capital rules effective November 2020, and UK Companies House requires the average number of employees as a mandatory disclosure that cannot be stripped out even from abridged small-company accounts. LinkedIn counts run consistently high because they count anyone who lists the employer, including leavers, contractors and advisers. The argument of the page: the total is the least useful part, and the three derived readings that matter are function mix (what kind of company they are), rate of change (trajectory), and geographic concentration (which market they are entering next). Cannot be found: current exact number, team-level breakdown, attrition, contractor and offshore ratio.
- [How to Find a Competitor's Tech Stack: 9 Sources, Compared](https://www.flares.tech/sources/competitor-tech-stack): Detection tools only fingerprint what the browser downloads, which covers two of five stack layers and not the ones that matter competitively. Includes a layer table (marketing/front end, third-party services, infrastructure, data and backend, internal tooling) mapping each to the sources that reveal it. The two under-used sources are the competitor's own subprocessor list — a maintained, public, company-authored inventory of hosting, email, support, analytics and payments vendors, published because enterprise contracts and data protection agreements require change notification — and six months of engineering job ads, which name the backend precisely because the role depends on it. Also covers view-source and the network tab, engineering blogs and public repositories, DNS records (MX, CNAME, TXT), docs and status pages, app store privacy labels, and review sites. Legal section draws a sharp line between reading what is published and probing systems you have no permission to test.
- [How to Find Competitor Ad Spend: 8 Sources and Why Every Number Differs](https://www.flares.tech/sources/competitor-ad-spend): States plainly that no public source reports a commercial advertiser's spend, and that every third-party estimate multiplies modelled traffic by a modelled cost per click, which is why two tools disagree by multiples. Includes a coverage table for the public ad libraries: Meta Ad Library (all active ads with creative and start date; spend ranges only for political and social-issue ads), Google Ads Transparency Center (verified advertiser and creatives across Search, YouTube and Display for roughly 30 days, no spend or impressions for commercial advertisers), LinkedIn Ad Library (active sponsored content, the B2B channel most articles omit, up to a year after last impression), and other EU Digital Services Act repositories. The three things you CAN get exactly: creative and cadence from the libraries, relative bidding pressure from your own Auction Insights report (a measurement, not an estimate), and an audited ceiling from filed sales and marketing expense. Also answers how to identify competitors advertising in ChatGPT: there is no public ad library because ads are served inside private one-to-one conversations, so the proxies are running buying-intent prompts yourself and watching competitors' hiring and agency announcements.

- [How to Find Competitor Market Share: 10 Sources and 3 Ways to Calculate It](https://www.flares.tech/sources/competitor-market-share): Argues that the hard half of market share is the denominator, not the numerator: measured market totals exist in regulated industries, consumer goods retail panels, government statistics and public procurement, and in most business software categories nothing is measured, so every circulating share figure is estimated revenue divided by an estimated market size. Gives three calculations and when each is supportable — absolute share (needs a measured total), relative market share (yours divided by the largest rival's, needs no market size at all, the axis of the growth-share matrix), and share of deals contested (win rate by named competitor from your own CRM, the only fully auditable one). Ten sources: analyst trackers, filings and earnings calls, industry regulators, government statistics agencies, library business databases (IBISWorld and Statista are free remotely with a public library card), retail measurement panels, public procurement awards, technology detection install counts, review and app store counts, and your own CRM. Also covers unit share versus revenue share, with revenue share divided by unit share giving a relative price index, and the Herfindahl-Hirschman Index for concentration. Legal section covers the antitrust exposure specific to this data point: exchanging current or forward-looking volumes with competitors, including through a trade association, is unlawful information exchange, while third-party aggregated historical benchmarking is the accepted form.
- [How to Find Competitor Funding: 10 Sources, Including Unannounced Rounds](https://www.flares.tech/sources/competitor-funding): Funding databases are built from announcements, so they are blind to any round a company chose not to publicise. Issuing shares is a mandatory registry filing instead. Includes a jurisdiction table: UK return of allotment of shares (form SH01) at Companies House within one month under section 555 of the Companies Act 2006, stating the amount paid per share including premium so the raise can be computed; US Form D at the SEC within 15 calendar days of first sale under Rule 503(a), which gives offering size and officers but no investors or valuation; German shareholder lists at the Handelsregister naming every shareholder and their holding, downloadable free; French capital increases registered with the commercial court and published in the official bulletin. Key caveat most guides omit: filing Form D is not a condition of the Regulation D exemption, and research by Qianqian Yu and Kathleen Weiss Hanley at Lehigh University finds more than half of venture-backed private firms never file, so an absent filing is no evidence. Also covers public grant and subsidy registers (the project abstract is a competitor describing their unreleased roadmap), equity crowdfunding archives (pitch decks, financials and pre-money valuation), and how to read a round: who led and at what stage, time since the last one, equity versus debt, and the hiring in the ninety days after.
- [How to Find Competitor Keywords: 10 Sources, Including Free Ones](https://www.flares.tech/sources/competitor-keywords): Reframes competitor keyword research as competitive intelligence rather than SEO reporting: a keyword set is a competitor stating which segments they chase, which comparisons they fear and which problems they claim. Makes the distinction the tool vendors do not: the keyword lists are close to observation and reasonably reliable, while every number attached (volume, traffic, difficulty) is a proprietary model, so figures must never be compared across two tools. The only measured competitor search data anyone has is free and first-party: Search Console queries with high impressions and a position outside the top five are terms where the market already considers you relevant and a competitor is taking the click, and Auction Insights names the advertisers in your auctions. Free route also covers Google Ads Keyword Planner (accounts with no spend see volume brackets rather than exact figures), Bing Webmaster Tools keyword research, autocomplete and People Also Ask, and the competitor's own sitemap, which shows the plan including sections not yet ranking. Legal section covers brand bidding: the Court of Justice of the European Union held in the Google France cases (2010) that the search engine does not infringe by selling the keyword, and in Interflora v Marks & Spencer (2011) set the test that an ad must let a reasonably well-informed and reasonably observant internet user tell the advertiser is a third party unconnected with the trademark owner. The exposure is the ad text, not the keyword.
- [How to Find Competitor Website Traffic: 10 Sources and How Accurate Each Is](https://www.flares.tech/sources/competitor-website-traffic): States that no external tool counts a competitor's visits and quantifies the error using SparkToro's comparison against real Google Analytics data, published by Rand Fishkin in November 2022 across 641 websites and 7,692 monthly comparisons: correlation with actual traffic was Semrush 0.790, Datos 0.720, Similarweb 0.659 and Ahrefs 0.504 (organic search only, so it understates total traffic by design), while Similarweb led on a different measure, landing within 30% of the real figure around 63% of the time. Errors of 100% or more were common and accuracy collapsed on small sites. The subtle finding: the provider ranking first on correlation is not the one ranking first on absolute accuracy, so "which traffic checker is best" has two different answers depending on whether you need direction or level. The differentiating technique is bounding any estimate against two free datasets that measured real activity — the Chrome User Experience Report's popularity rank magnitude, derived from real Chrome navigations and expressed in buckets from top 1,000 to top million, and Cloudflare Radar's domain rankings, built from 1.1.1.1 resolver data counting unique client networks rather than raw queries. Rule that falls out: never quote an absolute visit count, quote a ratio against your own traffic from one tool, because systematic bias largely cancels out of a ratio. Also covers reading the channel split rather than the total, and why the direct bucket absorbs unattributed traffic.

- [How to Find Competitor Partnerships: 10 Sources, Including the Contract](https://www.flares.tech/sources/competitor-partnerships): Built on two techniques the incumbent pages omit. First, research the partner rather than the competitor: every partnership has two sides and the junior party publicises it harder, so searching the competitor's name against marketplaces, partner newsrooms, certification registries and systems-integrator sites surfaces relationships the competitor never announced. Second, for a listed company the agreement itself is often on the public record — US disclosure rules require material contracts to be filed as exhibits and name partnership, joint-venture and similar agreements as qualifying, and entering a material definitive agreement triggers a current report normally within four business days; companies may redact commercially sensitive detail without asking the regulator first but the redactions must be clearly marked, so you can see exactly which terms both parties considered dangerous to publish. Includes a partnership-type table (technology/integration, channel/reseller, embedded or white-label, co-marketing, services) mapping what each commits, what a concentration signals and the threat level of each. Also covers finding ended partnerships by diffing the partner directory against archived versions, and disambiguates two contaminating intents: "4 types of partnerships" is a legal entity question (general, limited, LLP, LLLP) and the "5 D's of partnership" is a succession-planning framework with no canonical version.
- [How to Find Competitor Churn Rate: What You Can and Cannot Know](https://www.flares.tech/sources/competitor-churn): States plainly that no public source reports a competitor's churn rate, because churn is computed from a customer list held in one billing system and no regulator requires it. The key analytical point: net revenue retention, which listed software companies do disclose, nets expansion against losses, so a figure well above 100% is fully compatible with heavy logo churn underneath — gross revenue retention is the honest figure and is published less often. Includes as-reported retention figures with periods (Commvault SaaS NRR 122% at 31 March 2026 and 127% a year earlier; Intapp cloud NRR 120% trailing twelve months at 30 June 2025; Duck Creek SaaS net dollar retention 120%/117%/114% for fiscal 2021/2020/2019; Thryv seasoned NRR 98%/96% for 2024/2023; ON24 89%/82%/87% for 2024/2023/2022), noting that every company names and defines the metric differently and that a figure below 100% means the cohort is contracting. What you can observe instead is named departures: technology removal detection records sites that dropped a product (an observation, not a model), archived customer-page diffs list logos that vanished, and your own inbound from their customers is measured and first-party. Also covers reading retention behaviour (lock-in lengthening, customer success outgrowing sales, win-back campaigns appearing, the roadmap turning inward) and disambiguates employee churn/turnover, which dominates the search results and is a different metric.
- [How to Track Competitor Hiring: 10 Sources and What Each Role Reveals](https://www.flares.tech/sources/competitor-hiring): A job advert is a budget already approved, which makes hiring the earliest external signal a company emits and one it cannot spin. The practical trick: job sites carry a filtered subset because pushing a role externally is a separate paid decision, while the applicant tracking system hosting a company's board carries every live requisition — click any apply button on their careers page and the destination tells you which system they run, then bookmark that board instead. Includes a table decoding the five parts of a job description that carry information (responsibilities = the plan for the next two to four quarters, required tools = the stack committed to, "you will work closely with" = the org chart, seniority language = inventing versus scaling, location = which market is being staffed), and a pattern table reading hiring clusters, first-ever roles, geographic expansion, customer-success outgrowing sales, broad simultaneous hiring (usually an unannounced funding round), tail-offs and roles that appear then vanish unfilled (cancelled plans, only visible via archived pages). Distinguishes itself from competitor headcount explicitly: headcount is a stock taken from filings, hiring is a flow that leads by months. Disambiguates the 70/30 recruiting rule (hire at 70% of qualifications, widely repeated with no attributable origin), recruitment trackers and the "5 stages of the hiring process", all of which concern running your own hiring.
- [How to Find a Competitor's Product Roadmap: 11 Sources Ranked by Lead Time](https://www.flares.tech/sources/competitor-roadmap): Rejects the standard answer ("watch their changelog") on the grounds that a changelog is a record of the past, and organises every source by how far ahead of launch it becomes visible. The lead-time table runs from patent applications (published roughly 18 months after the earliest filing date under US law, so they disclose old work but the longest window; applicants who certify they will not file abroad can request non-publication, so absence proves nothing, and most patents never ship) through public grant awards (6-24 months), job postings for capabilities they do not ship (2-4 quarters), trademark filings (a product name becomes a searchable public record within days of filing, and intent-to-use filings exist precisely for unreleased products), investor commentary, conference abstracts, beta programmes, documentation and API reference diffs (days to weeks; new permission scopes are the most revealing because a scope must describe precisely what it grants and is added only when something needs it), the public roadmap (low reliability for timing, high for showing what customers are demanding via vote counts), the changelog (zero lead time, but the only honest measure of shipping pace) and acquisitions (immediate). Also gives a cost test for separating real roadmap items from marketing: rank findings by what they cost to produce — nothing (a slide), money or legal fees (a trademark or patent filing), people (specialist hires, an acquisition), or engineering time already spent (a new API endpoint, a documented beta, which means the thing exists).
- [How to Analyze Competitor Reviews: The Method, Tested on 500 Reviews](https://www.flares.tech/sources/competitor-reviews): The method rather than a tool list, demonstrated on Flares' own published study of 500 verified G2 reviews across four competitive intelligence tools coded into 340 distinct themes. Two rules make the difference. Apply a materiality threshold before anything is called a finding — three per cent of a given competitor's reviews is defensible, and in the study it cut 28 universally praised capabilities down to 7 genuine table stakes, and 11 shared complaints down to a single one clearing the bar against all four tools. And code at two levels, a specific theme (what a team can act on) plus a parent category (what shows five small complaints are one structural problem). Includes a platform comparison by verification model: G2 verifies identity on every review (LinkedIn profile, verified business email, or personal email plus a product screenshot, with a current-user badge for screenshot-validated reviews) and moderates by hand; Capterra uses human moderators plus automated authenticity and plagiarism checks; Trustpilot labels each review verified or unverified but has a lower verification rate for business software; app stores give velocity rather than substance. Covers what reviews reveal that a feature comparison cannot — including a ninefold spread in customer-success themes between two vendors (47% versus 5%), an operating-model difference no feature grid contains — and states the five structural limits of any review corpus (self-selection, absence is not evidence of absence, platform skew, temporal span, and the fact that reviews are written by users who stayed).
- [How to Analyze Competitor Positioning: 10 Sources and What Buyers Hear](https://www.flares.tech/sources/competitor-positioning): Splits competitor positioning into two datasets that most analyses collapse into one. The claimed position comes from surfaces the competitor controls (homepage and category word, pricing and packaging, their own comparison pages, paid search copy, product marketing job adverts); the received position comes from surfaces they do not (review text and category placement, third-party comparison pages, analyst evaluations, and your own win/loss interviews). The finding is the gap between them, written as one sentence. Extraction uses the five components April Dunford sets out in Obviously Awesome (competitive alternatives, unique attributes, value, target customer characteristics, market category), each mapped to where it is observable on somebody else's company. Four markers date a genuine repositioning as opposed to a campaign: the category word, the named proof logos, the entry pricing tier, and whether the change survives two archive snapshots. Also explains that G2's vertical Grid axis is Market Presence, which blends review counts with company-size and web-visibility signals, so a high placement is partly a measure of reach rather than quality. States plainly which frameworks are real (Trout 1969 and Ries & Trout 1981; Park, Jaworski and MacInnis in the Journal of Marketing 1986 for functional/symbolic/experiential; Youngme Moon in Harvard Business Review May 2005 for reverse, breakaway and stealth positioning; Porter's generic strategies and four corners) and which are invented for search traffic (the "4 types of positioning", the "5 P's of positioning", the "5 criteria for understanding competitors").
- [How to Find a Competitor's Go-to-Market Strategy: 10 Sources and 6 Tells](https://www.flares.tech/sources/competitor-go-to-market-strategy): Starts from the fact that many competitors have already written the answer down: a full-text search of the SEC archive, which covers filings from 2001 onwards, returns more than a thousand registration statements containing the exact phrase "our go-to-market strategy". Klaviyo's 2023 filing states its strategy "is primarily product-led", attracting most new customers "through inbound channels, such as word-of-mouth, agency partnerships, and platform integrations", and discloses a 14-month customer acquisition cost payback period and a quarter-by-quarter net revenue retention series from 111% to 119%; PagerDuty's 2019 filing describes self-service landing complemented by "a high-velocity inside sales team" for midmarket and SMB and "a field sales team focused on enterprise customers". For private competitors the motion is reconstructed from six observable tells, given as a table mapping each to self-serve, inside sales or field: entry price and minimum, signup gate, revenue-role ratio on the job board, channel and marketplace presence, case-study mix, and event sponsorship level. Tells are ranked by what it would cost to change them, so pricing and packaging beat marketing copy. Also covers running the competitor's own signup flow under your real identity as repeatable dated evidence, triangulating their segment from the pricing floor, integrations, compliance pages and testimonial job titles, and the six signals that a competitor is moving upmarket. Disambiguates the 3-3-3 rule in marketing (at least three incompatible versions, no attributable origin), the 4 Ps of GTM (the marketing mix, which describes an offer rather than a route to market) and the "5 pillars" of GTM.
- [How to Find Competitor Patents: 9 Sources and the Assignee Problem](https://www.flares.tech/sources/competitor-patents): Argues that the search box is the easy part and the owner name is where searches fail. Gives a table of the five gaps that make an owner-name search incomplete (trading name is not the legal name, subsidiaries file separately, acquired portfolios keep the old name, founders file personally, name variants and typographical errors) with the fix for each, including the fastest one: the terms of service and privacy policy must name the contracting legal entity precisely. Nine sources: Google Patents (120M+ publications, 100+ offices), USPTO Patent Public Search (which replaced four legacy tools at the end of September 2022), EPO Espacenet (worldwide coverage and the best free patent-family view), WIPO Patentscope, USPTO Patent Assignment Search (ownership transfers since 1980, plus security agreements showing patents pledged as debt collateral and their releases), patent registers and file wrappers (what an examiner forced the applicant to narrow), the competitor's own virtual patent marking page (a company-authored map of products to patent numbers, published to preserve damages), litigation and post-grant review dockets (a challenger served with an infringement complaint has one year to petition for review), and commercial analytics platforms. Explains which record fields carry intelligence: legal status first, since US maintenance fees fall due at 3.5, 7.5 and 11.5 years and an unpaid one drops the invention into the public domain early; then priority date, patent family, classification codes, citations, inventors, background, and claims last. The family is read as a paid statement of intended markets, because international applicants have around 30 months to choose which national offices to enter and each one costs fees and translation. The legal section is the only one in the cluster where the risk is created by reading: US courts may enhance damages for wilful infringement, and the Supreme Court's 2016 Halo Electronics v Pulse Electronics decision lowered the bar by removing an objective-recklessness requirement, so clearance work belongs with counsel rather than in a shared research document.
- [How to Find a Competitor's Leadership Team: 10 Sources With Dates](https://www.flares.tech/sources/competitor-leadership-team): The names take ten minutes and are worth little; the intelligence is in which role was created, who filled it and what they did before. Gives a jurisdiction table for where a change is officially dated: US listed companies file a current report on the departure or election of directors and the appointment of principal officers normally within four business days, including the appointee's business experience over the previous five years, with annual proxy materials covering board composition and executive compensation; UK companies must notify Companies House of a director appointment or termination within 14 days and the filing history and PSC register are free; Germany publishes managing directors and shareholder lists through the commercial register; France registers officer changes with the commercial court. Includes a table reading eight appointments for what their existence implies over the next two to four quarters (first CRO, first CPO, a CFO with public-company experience, first partnerships lead, a country general manager, a first security or compliance leader, a chief customer officer after a growth phase, and any interim or acting title) with what to check next for each. The prior company is treated as the playbook, since senior hires re-run what worked before, usually within two or three quarters. Departures get equal weight, read by what happened to the seat: replaced, absorbed by a colleague, removed entirely, or several exits in one quarter. The legal section is the only one in the cluster about personal data rather than trade secrets, covering professional context only, the European information duty that attaches to personal data you did not collect from the person, no pretexting, and keeping recruitment separate from research.
- [How to Check a Vendor's Security Certifications: 10 Sources](https://www.flares.tech/sources/vendor-security-certifications): Splits security frameworks into three tiers that decide what evidence you can expect. Publicly verifiable: ISO/IEC 27001 through the accreditation forum's global database (IAF CertSearch) or a national accreditation body, FedRAMP through its own marketplace with ready/in-process/authorised statuses, PCI DSS service providers through the card network's global registry with annual revalidation, CSA STAR (free, with the completed cloud questionnaire attached), and UK Cyber Essentials through the delivery partner's certificate search. Report-only: SOC 2 has no register anywhere because it is a restricted-use attestation, so the artefact is the report, and SOC 3 is the general-use version written for public distribution. Not a certification at all: HHS does not endorse or recognise private organisations' HIPAA certifications and holding one absolves nobody. Two checks precede everything: the standard version, since certificates naming the 2013 revision of ISO 27001 became invalid on 31 October 2025 after a three-year transition, and the scope statement, which often covers a head office rather than the product processing your data. Distinguishes certified, audited, compliant and aligned as four different claims. For competitive rather than supplier use, the certification list is a dated map of which markets a rival has decided to enter, with an in-process federal listing being a public commitment months ahead of authorisation. Legal section covers restricted-use reports received under agreement, not requesting evidence under a false purpose, and registers whose own terms restrict them to verification.
- [How to Analyze Competitor Social Media: 10 Sources and a Method](https://www.flares.tech/sources/competitor-social-media): The only source in the cluster with no archive behind it: posts fall out of the feed within days, ephemeral formats expire in 24 hours, deletions leave no trace, and web archives capture these pages badly, so the method is capture-first and the analysis you can run is bounded by the capture you started months ago. Ranks the surfaces by how managed they are, from the company page (most reviewed, least informative) through paid creative, executive accounts and employee posts to the replies and comments (least managed, most informative), and identifies employee posting volume around a product area as the most reliable early signal. Rejects engagement rate as a measure of the social team rather than the company, and replaces it with a seven-category post classification (announcement, named proof, commercial, recruitment, event, thought leadership, community) read as a mix shift over time. Free first-party tools: both large business platforms let a page admin add competitor pages inside native analytics and compare follower growth and organic engagement; public advertisement and commercial content libraries are searchable by advertiser and, for the largest platforms in the EU, legally required with roughly a year of retention after an ad last ran. Notes the 5:3:2 content rule has no primary source and concerns your own posting rather than competitor analysis. Legal section is the only one in the cluster about participating rather than reading: no invented identities, disclose your employer before engaging, and treat employee posts as personal data.
- [How to Find a Competitor's Average Deal Size: 10 Sources](https://www.flares.tech/sources/competitor-average-deal-size): Argues the average is the wrong number and proves it on disclosed figures: UiPath's annual report for the year ended 31 January 2026 shows revenue of $1,610.6m, ARR of $1,852.6m and approximately 10,747 customers, with 2,565 customers at $100k or more of ARR and 357 at $1m or more, accounting for about 89% and 52% of revenue. Revenue divided by customers gives roughly $150,000, while the three quarters of customers outside the $100k band share about 11% of revenue, or roughly $21,700 each, so the headline average sits close to seven times above what a typical customer represents. What listed competitors disclose is the distribution rather than an average, and the customer definition (subsidiaries counted once, free and trial users excluded) must be read before using any denominator. Public sector contracts publish whole deals with the buyer named, the value stated and the term attached: the US federal spending database covers awards above $25,000 with reliable data from 2008, the EU tender journal publishes contract award notices with winner and value, and UK regulated procurements begun after 24 February 2025 are published on the central digital platform under the Procurement Act 2023, with freedom of information requests reaching the contract itself. Includes an error-bar ranking of seven estimation methods, and a note that pay-transparency rules (including the EU directive whose 7 June 2026 transposition deadline only four member states met) make seller on-target earnings a weak sanity check. Legal section covers confidentiality clauses that bind the buyer rather than you, records requests being non-anonymous with supplier pre-disclosure notice, and never discussing commercial terms with a competitor.
- [How to Analyze Competitor Strategy: 10 Sources and Four Corners](https://www.flares.tech/sources/competitor-strategy): The umbrella page for the cluster, mapping ten observable layers (position, motion, price, deal shape, product direction, capability, fuel, distribution, scale, public voice) to the page that covers each. Uses Porter's four corners because its output is a prediction of competitor response rather than a profile: future goals from reported targets and executive compensation, current strategy from pricing and hiring, assumptions from the competition section and risk factors, capabilities from headcount, patents and balance sheet, with assumptions identified as the corner that produces surprises and the one nobody fills in. Signature technique is a year-on-year diff of the same filing section, demonstrated on HubSpot: the list of principal competitive factors ran to twelve items in the annual report for the year ended 31 December 2024 and fifteen for 2025, with all three additions about AI (agents handling end-to-end workflows, a unified data platform with AI-powered data quality, and answer-engine and AI-native marketing), while the competitor category list grew from ten to twelve, adding CPQ and billing providers and AI-native CRM and workflow automation startups. More than 8,500 annual reports in the SEC archive contain the exact phrase "our principal competitors". The private-company equivalent is the UK strategic report, required of any company not entitled to the small companies exemption and containing a fair review of the business and its principal risks. Contradictions are resolved by reversibility: a blog post takes an hour to undo, a price change a quarter, a senior hire a year, an acquisition never. Deliverable is a response profile table of your candidate moves against their likely reaction. Legal section is the only one where the risk is your own document rather than the material collected.
- [How to Use SEC Filings for Competitive Analysis](https://www.flares.tech/sources/sec-filings): The source manual the rest of the cluster points at. Covers eleven filing types with what each gives a competitive analyst and its deadline: the annual report on Form 10-K (business description, competition section, risk factors, exhibit index), the quarterly report on Form 10-Q, the current report on Form 8-K within about four business days, the registration statement on Form S-1, the proxy statement on Form DEF 14A (board and the performance measures bonuses actually pay against), Form 4 insider transactions within two business days, Schedules 13D and 13G with the initial 13D deadline now five business days, Form D exempt offering notices, Forms 20-F and 6-K for non-US listed competitors, fund portfolio reports carrying an investor's own fair value mark on private shares, and exhibits containing material contracts and the list of subsidiaries. Prescribes reading order: business, competition, risk factors, management discussion, exhibit index, financial statements last. The signature technique answers the standard objection that the competitor is private: full-text search runs across everybody else's documents, and Databricks, which has never filed an annual report, is named in 8,976 filings by other companies, broken down as 7,159 fund portfolio reports, 118 quarterly reports, 110 annual reports, 101 current reports and 41 exempt offering notices, with one listed rival naming it in seven separate annual reports. Access notes that full-text search covers 2001 onward and caps its result count, so a common phrase reports a ceiling rather than a total. Includes a misreadings table (a competition section is a defensive disclosure rather than a market map; segment revenue is not product revenue; an earnings release is furnished under a lighter standard than the report that follows) and covers non-reliance filings, which are a competitor stating on the record that numbers they already published were wrong.
- [How to Use Earnings Calls for Competitive Intelligence](https://www.flares.tech/sources/earnings-calls): Treats the quarterly call as two documents with different evidentiary weight. The prepared remarks are drafted, legally reviewed and rehearsed, so a changed word or a dropped metric is a decision; the analyst question-and-answer session is unscripted and is the only recurring event where a competitor's leadership answers questions it did not choose, in public, on a known schedule. Signature technique is the category read: sell-side analysts are organised by sector, so the same small group covers every listed company in a market and asks each of them the same question inside a six-week reporting window, which turns one afternoon of transcripts into a comparative survey with every chief executive answering on the record. Also covers the indirect route to a private competitor by listening to the calls of their listed customers, partners, resellers and rivals. Eleven components with what each yields: prepared remarks, analyst Q&A, guidance, the earnings press release furnished the same day, the investor deck, transcripts, the replay (typically retained around 30 days), the quarterly report filed days later, investor days, sell-side fireside chats, and third-party calls. States plainly that guidance is a negotiating position rather than a forecast, that tone analysis is over-sold because executives are media-trained, and that a competitor naming you is usually a prepared framing rather than evidence you are winning. Notes that over 800 current reports on the regulator's database contain the phrase "earnings call transcript", so some transcripts are permanently archived in a public system. Says plainly that the "7 Ps of competitive intelligence" is not a real framework and is the extended marketing mix.
- [How to Use Review Sites for Competitive Analysis](https://www.flares.tech/sources/review-sites): Reframes a software review site as a database of buyers in which the opinions are one column. The highest-value surface is the reviewer firmographic filter, which segments a competitor's reviews by company size, industry, role and region and therefore produces a free readout of who actually buys from them; building that distribution for yourself and two rivals is a segmentation map derived from buyer behaviour rather than from a workshop. Documents the 2026 consolidation and why it weakens cross-platform corroboration: G2 announced on 29 January 2026 an agreement to acquire Capterra, Software Advice and GetApp from Gartner, putting the combined footprint at roughly six million verified reviews and more than 200 million annual software buyers, and Gartner's own quarterly filing records completion of the sale of its Digital Markets business in February 2026 for approximately $104.8 million net of cash transferred, after a $150.0 million goodwill impairment against that unit in the third quarter of 2025. Gartner Peer Insights was not part of the transaction. Covers eleven surfaces including feature-level scores, comparison and alternatives pages as a buyer-defined competitive set, and vendor replies as the only field a competitor writes themselves. Explains that the vertical axis of the best-known category quadrant is market presence, which blends review counts with employee data, web-authority scores and search volume, so it substantially measures size; and that G2's own documentation states its pay-per-click product has no impact on organic ranking or market report inclusion while sponsored listings do appear on category, alternative and discussion pages. Legal section is about what you may write rather than what you may read: never review a competitor under any identity, never approach anonymous reviewers.
- [How to Use Crunchbase for Competitive Analysis, and Its Limits](https://www.flares.tech/sources/crunchbase): Starts from how the data gets there, which decides whether a gap means anything. Crunchbase's own published description names four channels: direct input from companies, investors and partners including more than 4,000 venture partners and 600,000 active contributors; aggregated anonymised research behaviour from more than 80 million users; more than 1,000 news outlets and government filings; and internal analysts applying automated grading with human validation, producing more than 30 million verified updates a year. The consequence is that coverage follows incentive rather than importance, so a sparse profile is evidence about a company's fundraising and publicity habits rather than about its size or threat, and bootstrapped or non-US competitors are systematically under-described. Second technique is a field-by-field split between recorded events (funding rounds, investors, acquisitions) and model outputs (employee bands, revenue estimates, growth scores, similar-company lists, and the weekly-refreshed prediction scores the vendor publishes its own unaudited accuracy claims for), which the profile styles identically. Covers eleven field groups, the free tier's profile-view cap and absence of exports and alerts, the mid tier's published limit of up to 2,000 export rows a month, and a comparison table against PitchBook and CB Insights, Tracxn and Dealroom, national company registers and the competitor's own announcements. Misreadings covered: total funding measures capital consumed rather than strength, the last round is not a current valuation, a Series B label is a name the company chose, and the similar-companies list is a classification output rather than your competitive set.
- [How to Find What Technology a Website Uses: 11 Methods](https://www.flares.tech/sources/website-technology): Splits one question into five that have different answers: the CMS or site builder, the front-end framework, the hosting and CDN, the third-party services, and the server-side language. Four are settleable from a browser and the fifth generally is not, demonstrated with an original measurement: reading the response headers of thirty well-known business software home pages on 7 August 2026, twenty-seven returned a server header and every one named a CDN, edge platform or web server rather than a programming language, three returned none, an x-powered-by header appeared eight times and named the same front-end framework on all eight, and a generator meta tag appeared on five of the thirty. Eleven methods covered one H3 each: lookup tools, the page source and generator tag, developer tools, response headers, platform-specific paths and well-known files, robots.txt and the sitemap, public DNS records, the TLS certificate and public certificate transparency logs, public repositories and package manifests, archived copies of the same pages, and aggregate surveys. Includes a question-to-method routing table, a dedicated WordPress detection section anchored on W3Techs figures for 7 August 2026 (41.2% of all websites, 59.1% of sites using a content system, 30.4% using none of those monitored) with the headless caveat that makes a negative result weak evidence, and a section on why fingerprint-based checkers are reliable on presence, blind to self-hosted and proxied vendors, and slow to notice removals. Boundary with the tech-stack page is stated explicitly: that page maps a company's whole stack, this one identifies what one website is built with.
- [How to Use Reddit for Competitor Research: Method and Limits](https://www.flares.tech/sources/reddit): Treats the vote as the distinguishing property. Reddit's own annual report for 2025 states that every post and comment starts with a single upvote and must earn visibility through community members boosting it, that voting in both directions is anonymous, and that moderation depends on volunteers, which makes a score an unpaid judgement by the audience about whether a criticism resonates, something no review corpus produces. The same filing supplies the counterweight, describing accounts automated to mimic posting, commenting and voting and noting that language models make detection harder, so a score measures resonance and never truth, scale or correctness. Scale figures from the same source: over two billion posts and over 22 billion comments through 31 December 2025, more than 100,000 active communities, and an average of 121.4 million daily active uniques in the fourth quarter. Second wedge covers Reddit as a corpus AI answers are assembled from, which makes reading the category's threads a preview of what a model will tell the next buyer. Eleven surfaces one H3 each, plus a six-row search table (general web search restricted to the site beats the platform's own, community and author filters, the vocabulary of switching and pricing threads), a misreadings table, the public content policy line between non-commercial reading and commercial reuse, and the disclosure and no-pretexting rules for participation.
- [How to Use Job Postings for Competitive Intelligence](https://www.flares.tech/sources/job-postings): Argues that a job advert is the only competitor document penalised for vagueness, with a table ranking five published document types by what imprecision costs the company: home page and press release (nothing), analyst briefing (little), documentation (support tickets, and only about what shipped), job advert (wrong applicants, wasted recruiter weeks, a failed hire). Eleven parts of an advert covered one H3 each: title and level, the about-us boilerplate, responsibilities, requirements, reporting line and team, pay range and variable structure, location and sponsorship, success measures, the described hiring process, benefits and compliance statements, and posting metadata including the requisition identifier that distinguishes a new role from a repost. Two techniques nobody writes: the about-us paragraph is a dated positioning statement written by marketing and pasted into every advert, so a year of adverts is a free time series of a competitor's pitch; and because pay-disclosure rules attach to where a role sits, a company hiring the same title in several places publishes the band once, and that band describes the role rather than the location, with band width, base-to-variable split and movement between two versions each carrying separate meaning. Boundary with the hiring page is explicit: that one counts roles and reads the trend, this one reads the document. Legal section rules out applying under false pretences and separates competitive research from recruitment.
- [How to See a Competitor's Old Website: Web Archives, Ranked](https://www.flares.tech/sources/web-archives): The one competitive source that produces evidence a sceptic can check without trusting the researcher, because a capture sits at a public address containing its own timestamp on a third party's servers. Covers the Wayback Machine calendar and its side-by-side changes comparison, Save Page Now, archive.today's on-request rendering with a full browser, the retirement of search engine caches in 2024 and the Wayback link added inside Google's about-this-page panel in September 2024, surviving URLs and redirects on the live site, the company's own changelog and press archive, app store version history, national library web archives under legal deposit, curated institutional collections that are searchable by the full text of captured pages, and public repository history. Two limits stated as rules: an archive samples a website rather than recording it, so a snapshot proves what a page said on one date and never what it did not say between two captures, which is why findings should be reported as intervals; and archived pages fail to replay for five identifiable reasons, with an empty pricing table almost always meaning the prices were assembled in the browser rather than removed, recoverable by reading the snapshot's source. The Internet Archive passed one trillion pages in October 2025 and stopped treating robots files as an archiving instruction in 2017, so sites currently blocking crawlers often still have years of history.
- [How to Use Glassdoor for Competitive Intelligence: 11 Signals](https://www.flares.tech/sources/glassdoor): The only competitive source whose operator has published a measurement of its own sampling bias, which is what makes the "just angry ex-employees" objection answerable rather than arguable. Glassdoor's own economists Andrew Chamberlain and Morgan Smart published Give to Get: A Mechanism to Reduce Bias in Online Reviews in October 2017, covering more than 116,000 US company reviews from 2013 to 2016: the access rule cut one-star reviews by 3.6 percentage points and five-star by 2.1, while three-star rose 2.6 points and four-star 2.9. Eleven surfaces covered one H3 each: company reviews with their pros, cons and advice-to-management fields, the overall rating and its six sub-ratings, CEO approval and business outlook, interview reviews, salary reports, benefits reviews, the employer's own replies, the profile fields, job listings on the same page, the annual employees'-choice lists, and the employer-side analytics that benchmark sentiment against named competitors. Reads senior management and career opportunities as the two sub-ratings that move first and predict departures. Access section covers the give-to-get rule (one contribution buys twelve months), the Indeed account requirement from 18 November 2025, the 20 April 2026 linking deadline and the legal merger on 1 July 2026. Legal section is about individuals rather than trade secrets: never de-anonymise a reviewer, never quote one externally, and United States v. Glassdoor (Ninth Circuit, 8 November 2017) held the First Amendment did not shield reviewer identities from a grand jury subpoena.
- [How to Analyze Competitor App Reviews: Version by Version](https://www.flares.tech/sources/app-store-reviews): The only review corpus where every entry is bound to a build number and the vendor publishes dated release notes beside it, which turns sentiment into an engineering timeline you can date to the week. Establishes that a star rating is not a property of the app: Apple documents that the summary rating is specific to each territory and can be reset by the developer on a new version (applying to all territories at once, irreversible, written reviews survive, with a notice shown on the page); Google announced on 23 August 2021 that phone users would see ratings computed from their own country's reviews from November 2021 and device-specific ratings from early 2022, weighted toward recent ratings, and committed to warning developers of any shift above 0.2 stars ten weeks ahead. Third mechanism: Apple caps in-app rating prompts at three times in a 365-day period, so a rating gap between two apps can be a prompting-strategy gap rather than a quality one. Eleven fields covered one H3 each, including the histogram, version history, developer replies, country storefronts, the machine-written review summaries introduced in iOS 18.4, the sort controls, the developer's other apps, and third-party rating-history trackers. Operating rule: record storefront, date, app version and rating count with every figure.
- [How to See Competitor Ads: Every Public Ad Library, Ranked](https://www.flares.tech/sources/competitor-ad-libraries): Ad libraries are compliance artefacts built to a legal specification rather than research products, which explains the fields, the borders and the wildly varying quality. EU rules oblige very large platforms to publish a searchable repository covering ad content, advertiser, payer, run period, targeting parameters and reach, retained one year after the last impression. On 5 December 2025 the European Commission fined X 120 million euros in the first non-compliance decision under the Digital Services Act, citing a repository with "design features and access barriers, such as excessive delays in processing" that "lacks critical information, such as the content and topic of the advertisement, as well as the legal entity paying for it"; on 15 July 2026 it accepted a corrective plan adding content and targeting filters, in-page results, a response time cut from 200 seconds, full creative and destination addresses, and programmatic access. Retention is set by ad category, not platform: commercial ads vanish when campaigns end, EU-exposed ads last a year, political ads run to seven, and Google removed its entire EU political archive covering 27 countries back to 2018, spotted 28 September 2025, before Regulation (EU) 2024/900 applied on 10 October 2025. Eleven libraries covered one H3 each, ending with advertising inside AI assistants, where no repository exists at all. Method: search the legal entity not the brand, collapse the set into distinct ideas rather than counting ads, and rank by how long each has run, which is the only performance signal these libraries support.

- [How to Use Win/Loss Interviews for Competitive Intelligence](https://www.flares.tech/sources/win-loss-interviews): The only competitive research whose respondent personally sat through your pitch and a rival's, compared them under real budget pressure, and then chose. Covers the eleven conversations a programme can draw on (loss, win, no-decision, churn/switcher, evaluator versus economic buyer, the recording, the buyer's own scorecard or RFP, the rep debrief kept separate as a control, what buyers volunteer about a rival's commercials, the reference calls they took, and third-party programmes), each graded for reliability. Two arguments incumbent pages miss: seller and buyer accounts of the same deal diverge systematically, which Corporate Visions puts at 50 to 70 per cent of the time across what it describes as over 100,000 B2B deals, so the gap is the finding rather than a reconciliation problem; and win interviews are the half almost nobody runs, despite being the only place a competitor's live pitch against you is recorded by its audience. Also covers who should conduct interviews (never the deal owner), the 30-day window in which competitive detail decays, incentives, consent and recording, no-decision deals as their own category, and the five things interviews can never establish. Links to the free win/loss interview questions tool rather than restating a question bank.
- [How to Use Sales Call Recordings for Competitive Intelligence](https://www.flares.tech/sources/sales-call-recordings): A buyer objection is a paraphrase of a competitor's pitch, carried to you by the person it was aimed at, which makes your own call archive the nearest legitimate route to a rival's live positioning. Grades eleven parts of the corpus (discovery, demo, pricing and negotiation calls, the final call on a lost deal, renewal and account reviews, tagged objections, your own reps' answers, the transcript index, saved competitor alerts, scorecards and rep summaries). Second wedge nobody writes: half of what a competitor search returns is your own team talking, and the distance between the best answer anybody gave to an objection and the worst one in regular use is the real battlecard. Documents the failure mode that makes most archives look empty — transcription mishears short brand names, so a search returning nothing usually means the index lacks your spelling. Legal section covers the one-party federal consent baseline under the US Wiretap Act, the roughly dozen all-party states whose published lists disagree at the margins, GDPR notice and retention, and the rule that nothing leaves the company. Blind spot stated plainly: the archive holds only deals you were invited into.
- [How to Use CRM Data for Competitive Intelligence](https://www.flares.tech/sources/crm-data): The reason a CRM contains no competitive intelligence is structural rather than cultural. HubSpot's documented default deal properties include a closed lost reason and no property naming the competitor who won; Salesforce ships a standard OpportunityCompetitor object that has to be enabled and populated. The competitive layer is opt-in in both. Sets out the two-field rule (who won, kept separate from the stated cause), why the field must be required at the closing stage rather than at deal creation, and how to backfill four quarters from deal notes and recorded calls. Produces four numbers per named rival: win rate, median cycle length, median discount conceded, and average deal size in contested deals — with the discount gap between contested and uncontested deals identified as the most persuasive competitive figure available, because it prices a named competitor in your own currency from records the finance team maintains. Grades eleven CRM fields, separates no-decision outcomes from competitive losses, treats loss reason codes as evidence of what sellers believe rather than why buyers decided, and lists what should never be written into a CRM.
- [How to Use Company Registries for Competitive Intelligence](https://www.flares.tech/sources/company-registries): A registry filing is the only competitor document produced under legal compulsion, which is why it beats their own marketing on accuracy and loses to it on currency. Grades eleven filings (annual accounts, confirmation statement, incorporation documents, officer filings, share allotments, the charges register, beneficial ownership, insolvency notices, the official gazette, branch registrations, and bulk data or an API). Three things most guides get wrong: Companies House states that you can download company information for free, so the widespread claim that documents must be purchased is out of date; a UK private company has nine months after its year end to file and 21 months from incorporation for its first accounts, so entirely compliant figures can be nearly two years old; and Delaware, where a large share of software companies incorporate, states in its own guidance that its entity search returns only name, file number, incorporation date, registered agent details and entity type, with shareholder information not held by the Division at all and LLCs not required to list members or managers. Also covers the November 2022 CJEU judgment in joined cases C-37/20 and C-601/20 that invalidated public access to EU beneficial ownership registers, the legitimate-interest regime replacing it from 10 July 2027 under Regulation (EU) 2024/1624, and the April 2028 UK change requiring small companies to file profit and loss accounts while allowing them to opt out of publishing that information.
- [How to Use Competitor Press Releases for Competitive Intelligence](https://www.flares.tech/sources/press-releases): Press releases are weak evidence of what happened and strong evidence of what a company wants believed, and conflating the two is where competitor monitoring usually goes wrong. For a listed rival an announcement is a compliance instrument rather than marketing: under the SEC's Regulation FD a company may make material information public by a method reasonably designed to provide broad, non-exclusionary distribution, which is what a widely distributed release is, so timing and wording are constrained and there is normally a filed twin written under legal review worth reading alongside it. Supplies a five-part absence test — a named customer, a number with a unit, a specific date, a defined scope, a named executive owning the outcome — and treats what is missing as the finding. Grades eleven components including the About boilerplate as a silently dated positioning statement, the customer quote as a disclosed reference account, the decision to pay for wire distribution as a signal, and the announcement never issued. Notes that newsrooms prune old releases while investor relations archives retain them.
- [How to Use G2 for Competitive Analysis](https://www.flares.tech/sources/g2): Guides to this platform are overwhelmingly written for sellers with a paid dashboard; the public profile of any competitor is readable without an account and holds most of what a researcher needs. The central correction is that the headline score is a weighted model rather than an average. G2 documents that a review starts losing weight the moment it is submitted, recalculated daily, decaying gradually across the first 90 days, holding stronger weight for roughly 18 months, then accelerating until at about three years it carries approximately 3 per cent of its original weight in perpetuity. So a rival's score can fall with no new negative review, and the visible corpus is not the scored corpus, since reviews flagged for a conflict of interest can stay published while being excluded from scoring. The second axis is barely about reviews at all: G2's documentation enumerates market presence as company size, employee count, web presence, social following, website traffic, geographic presence, category-specific product visibility and the number of reviews, so only the last of the eight comes from customers, and several of the others cannot be observed from outside a company and have to be modelled. Grid eligibility is a volume threshold (a category needs at least six products with 10+ reviews and 150+ reviews overall; a product needs at least 10 in that category), so absence from a chart is evidence about review collection rather than about quality. The seller-side Competitors tab tracks up to five rivals and only within a shared category, which makes it a monitor for a competitive set already defined rather than a way to discover one.
- [Using Capterra for Competitor Research](https://www.flares.tech/sources/capterra): Position in a category listing is partly sold and the platform says so, which makes the paid layer readable evidence about competitor spending rather than a reason to distrust the site. Placements are bought by category rather than by search term, so the categories a rival sponsors are a public, dated statement of which buyer intents they consider worth money; Capterra states sponsored profiles may appear first, that buyers can identify them from the link-out button wording, that sorting and filtering are provided, and that compensation for traffic or leads does not affect its methodology, reviews, verification or ranking reports. The Shortlist methodology is published: products need at least 20 unique reviews on Capterra within 24 months of the research starting, lists carry up to 25 products and no fewer than 5 under the methodology current in January 2026, and both axes scale to 50 — ratings from reviews in the 24-month window, and popularity built from average monthly search volume, search engine result position, and review count and recency. That makes a Capterra position substantially a measurement of search performance, and explains why a product ranks differently here than on charts whose second axis measures company size. Also covers the catalogue breadth (Capterra states more than 100,000 providers) as the best free instrument for discovering competitors you did not know existed, the vendor-asserted feature checklist as the fastest category-wide feature map, and review verification (30+ QA moderators, 20+ control checks, incentives paid on approval regardless of rating).
- [Using Trustpilot for Competitor Analysis](https://www.flares.tech/sources/trustpilot): Every claimed Trustpilot profile has a public companion page, requiring no account, that reports twelve months of the business's own conduct rather than its customers' opinions: when the profile was claimed and what was verified, review counts split into organic, verified, invited and redirected, the star distribution broken out by review source, how many negative reviews the business replied to and when it last did, and how many reviews it flagged, at which ratings, and how those cases resolved including how many were found not to breach the guidelines. Because the star distribution is published per source, a competitor's invitation bias is measurable rather than assumed: subtract the organic rating from the solicited one and the gap is that rival's number. Trustpilot states invited reviews face the same fraud detection as any other, that consumers can review with or without an invitation, and that its rules prohibit influencing what reviewers say, incentivising reviews, or inviting only customers expected to be happy. It also publishes its own contamination rate: the Trust Report of 29 May 2025 reported removing 4.5 million fake reviews in 2024, 7.4 per cent of submissions, 90 per cent of them automatically and 53 per cent more automatically than in 2023, against 6.1 per cent in 2023, on a platform of over 300 million reviews with 61 million published in 2024. The TrustScore is a computed figure whose formula is not published, unlike the software directories' methodologies, and the behavioural record covers a rolling twelve months and then forgets, which makes a dated log the only long-term history.
- [How to Use Gong for Competitive Intelligence](https://www.flares.tech/sources/gong): Two separate mechanisms inside this platform answer to the word competitor and they do not count the same events, which is the correction the page is built around. Gong documents a tracker of its own named Competition that uses AI to detect competitive topics as a concept and works without a configured list of competitor names, alongside a keyword tracker named Competitors that an admin maintains, holds specific names, and is the tracker the competitor mentions insight in win/loss analytics reads. So the competitive analytics can only see rivals somebody remembered to type in, while the concept tracker sees the conversation regardless, and the difference between the two counts over the same period is an original, repeatable measurement of how stale a competitive set has become: filter by each, read the surplus calls, and harvest the names, the non-vendor alternatives such as spreadsheets and in-house builds, and the false positives. Keyword matching is verbatim unless the related word forms option is enabled, keyword trackers are unlimited in number, past calls are searchable with a tracker so a name added today returns its whole history, and Gong suggests no more than six trackers on an initiative board. The win rate on closed deals where a rival was named is a correlation inside your own funnel rather than a measure of the rival, because deals in which anyone names an alternative are larger, longer and more contested. Access facts: retention is the lesser of three years and the length of the customer relationship, shortening it purges irrecoverably, transcription covers 70+ languages while some analysis features return results only for English calls and example-based smart trackers are among them, the API documents three requests a second and ten thousand a day, and Gong publishes no pricing at all.
- [Using Modjo for Competitor Analysis](https://www.flares.tech/sources/modjo): Competitor detection here runs on topics, and Modjo's documentation states that the keywords you add are attached to a language and will only be detected in meetings held in that language. Everything competitive follows from that sentence: a multi-country archive is not one measurement but one measurement per configured language plus silence everywhere else, and silence renders identically to an absence of competition, so a team that built its topics for one market will read a genuinely contested second market as quiet indefinitely. The page sets out what equivalence between two markets requires (the spoken form of a name, the transcribed form, and matched effort on the competitive set) and treats opening a country as a competitive configuration task. Supplied topics covering competition, objections and product feedback ship in French, English, German, Spanish and Dutch; the custom vocabulary is the setting that decides whether short brand names are findable at all, and because corrections apply going forward the mangled spellings belong in the topic so the back catalogue stays searchable. AI Insights returns competitor names with mention counts and the outcomes of the deals attached, segmentable by deal size, industry, region and product through a connected Salesforce, HubSpot or Dynamics, with won and lost splits; the product itself displays a caution when the leading categories carry fewer than ten mentions, which the page adopts as a reporting floor because below it the ranking of two rivals inverts on a single deal. CRM filling writes the named competitor into a field from a per-field prompt with optional rep validation. Modjo publishes no prices and gates the insight layer by contracted plan.
- [How to Use Claap for Competitor Analysis](https://www.flares.tech/sources/claap): This platform labels competitor mentions and comparisons on every call alongside objections and concerns, pricing and negotiation discussion, and feature requests, applied as calls are processed rather than matched against a list somebody maintains. That inverts the usual trade: the archive can surface a substitute, a regional player or an in-house build nobody would have searched for, and it cannot show why any given label was applied, so recall is good and precision is unknown until you sample it. The method that follows is calibration before counting (read twenty to thirty labelled calls, establish what the label treats as a competitor, and check it catches the mentions you already knew), keeping a running list of unexpected names, and promoting a recurring name into a tracked competitive set with an owner. Coverage differs too: stand-alone recording through a browser extension and desktop application captures in-person meetings and uploads, which brings the late-stage and on-site rooms a scheduled bot never attends into the archive, and transcription covers 99+ languages with speaker identification. Retention is published per plan (three months on the free tier, two years, three years, unbounded at the top), which makes the ceiling on any competitive trend a purchasing decision you can look up rather than a fact nobody knows. A hosted MCP server, free to Claap users, searches transcripts by keyword or meaning with filters for company, contact, deal, folder, label, date range and the competitor, objection and pain point tags, returning matching passages grouped by recording; its deal tools require HubSpot and do not support Salesforce. Legal section covers the two questions specific to it: an in-person recorder gives no participant-list notice, and connecting an external assistant sends customer transcripts to a third-party model provider.
- [How to Use HubSpot for Competitive Intelligence](https://www.flares.tech/sources/hubspot): Two documented facts decide everything here and they point in opposite directions. HubSpot's default deal properties include a closed lost reason and a closed won reason and no property naming the competitor who won, so the reason a portal holds no competitive intelligence is that the form never asked for any. Meanwhile a family of properties has been filling itself in continuously on every deal the portal has ever run: date entered and date exited each pipeline stage, cumulative time in a stage, latest time in a stage since it was last entered, time in current stage, and days to close. The consequence nobody writes: one property added today makes four quarters of dormant history readable, because the timing it becomes readable against was recorded automatically while nobody was recording competitors. The page sets out how to read the stage that stretches (discovery means positioning, evaluation means a product comparison, negotiation means you are moving a price), why a median time in stage is trustworthy on far fewer deals than a win rate needs, and how the dropdown-select versus multiple-checkboxes choice decides whether later percentages sum. Association labels are covered as the route to holding a rival as a company record, documented at up to fifty labels per object pair, Super Admin to create, Professional or Enterprise for custom ones, with the marketing-audience exclusion named as the risk unique to running this inside a marketing platform. Access facts: exports in CSV, XLSX and XLS with a choice of properties and an emailed link, up to 300 exports in a rolling 24 hours and three at once; the custom report builder joins up to five sources and is Professional and Enterprise; calculation properties likewise.
- [Using Salesforce for Competitor Analysis](https://www.flares.tech/sources/salesforce): The page is built on a documented tension nobody else writes about. Salesforce's object reference describes OpportunityCompetitor as representing a competitor on an opportunity, holding a competitor name, a required link to the opportunity, and Strengths and Weaknesses fields capped at 1,000 characters each, with the documented purpose of associating multiple competitors with one deal and specifying the strengths and weaknesses of each. Its name field is typed as a combobox, which the same document defines as a picklist that also allows users to type a value not already in the list, so the field that presents as a controlled list accepts anything and produces three spellings of one company. Salesforce's own competitor-tracking training walks past the object, tells admins to build a restricted Competitor picklist plus a separate Lost Reason picklist, and makes the same argument about plain text using the Acme, Acme Inc and Acme Industries example. That module also specifies the validation rules (Competitor required at Proposal/Price Quote, Negotiation/Review, Closed Won and Closed Lost; Lost Reason at Closed Lost) and three reports assembled into one Competitive Analysis dashboard: opportunities by competitor and stage, lost reasons by competitor, and win or loss rates by competitor with formula columns. The signature section treats the Strengths and Weaknesses text as a longitudinal battlecard corpus, two thousand characters per rival per deal, and gives the reading method: the recurring sentence is the rival's real pitch, the claim present in wins and absent from losses is the objection you already answer, and the claim present only in losses is the one nobody has an answer for. Also covers the related list having to be on the page layout per record type, formatted versus details-only exports, and why no integration will fill an empty field.
- [How to Use Pipedrive for Competitor Research](https://www.flares.tech/sources/pipedrive): Pipedrive is the only major CRM that asks the competitive question without being configured to: marking a deal lost triggers a prompt for the reason, which is why the field is populated far more often here than its equivalent elsewhere. Pipedrive's documentation states that freeform is the default for that prompt, so a well-populated account can hold years of sentences nobody can count, and predefined lost reasons set by an admin in company settings are the twenty-minute fix. The setting that allows freeform alongside the predefined list is treated as a genuine trade rather than an oversight: it preserves the loss that fits no category, which is where market change appears first, and it also gives everybody a route past the list, so the page recommends keeping it on, reading those entries quarterly and watching the share that took it. Field choice is covered against Pipedrive's sixteen documented custom field types: Single option and Multiple options each hold up to a thousand predefined values, and an Organization-type field makes the rival a real record at the cost of sitting among your customers. The decisive access fact, quoted from Pipedrive's own documentation, is that only Premium and higher plans can use custom fields for filtering and building reports in the visual builder covering measure by, group by and segment by, while all users can use custom fields in the table view, which moves the analysis on the lower plans into an export. Required fields can be scoped to selected pipelines and stages, and Pipedrive states they will not stop deals progressing when updated via import, bulk editing, the API, automations or third-party integrations, which is why the page makes the measured coverage rate a permanent part of every report. Plan limits: custom fields 30/100/300/500 and Insights reports per user 15/50/250/500 across Lite, Growth, Premium and Ultimate, the 2025 renaming of Essential, Advanced, Professional and Power, and Enterprise. Exports cover deals with default, custom and system fields in Excel or CSV, are scoped to the exporter's visibility, and stay available for 28 days.
- [Using Attio for Competitive Analysis](https://www.flares.tech/sources/attio): The rare CRM where a competitor can be a record in its own right rather than a string inside a deal property, and the page follows that single difference through every section. Attio's data model is objects, records, attributes and lists; People and Companies are enabled by default, Deals, Workspaces and Users can be switched on, and custom objects are documented as Pro and Enterprise only against a total object allowance of three on free, five on Plus, twelve on Pro and unlimited on Enterprise. The join is a record reference attribute, documented as pointing to records of the same or another object and as bidirectional, so updating either side updates the other and a rival named on forty deals accumulates forty links with nobody maintaining a list. Company records enrich themselves: Attio documents name, logo, description, categories, location, estimated ARR, employee range, funding raised, social profiles, foundation date, team and follower count, with the last three of the money-and-size attributes excluded on the free plan. The page splits those into observed versus modelled, treats estimated ARR as a sorting key rather than a figure because the method is undocumented and unreproducible, and reads a sparse enrichment result as evidence about a rival's publicity habits rather than their size. AI attributes are covered from Attio's own documentation: summarise, classify, prompt completion, and a web agent that researches external sources in real time and returns reasoning and the sources used, at ten credits per record against one for the others, and available on all plans with credit limits by plan. The limit that shapes the workflow is that AI attribute values must be generated manually and do not run automatically. And the export finding that inverts the usual advice: Attio states enriched data cannot be exported and those cells will be blank in the file, so the most complete-looking view is the one that travels worst, and anything that has to leave must be typed into an attribute of your own.
- [How to Use BuiltWith for Competitor Analysis](https://www.flares.tech/sources/builtwith): Treats BuiltWith as a dated archive rather than a technology checker, which is the distinction the rest of the internet misses. Every technology on a profile carries the month it was first detected and the month it was last found, so the source answers when a competitor changed something rather than what they run today, and it is the only free way to see what a rival has STOPPED using, because a fingerprint signature only fires on something still present. Original measurement, repeatable by anyone: the free detailed profiles of five business software domains read on 6 September 2026 carried 234, 373, 330, 440 and 378 rows with a date pair, of which 113, 80, 69, 192 and 136 respectively were marked as no longer detected, so 590 of 1,755 rows (about a third) were a record of departures. The free tier withholds NAMES rather than data: it named exactly twenty-one technologies on each of the five domains and replaced the rest with an upgrade prompt while still printing each withheld row's full category tags and both dates, which is frequently enough to identify the vendor unaided. Covers reading the interval as a lifespan (embedded, failed experiment, new capability, migration in progress, where two adjacent rows in one category date a platform switch to the month), the relationship profile (BuiltWith states it tracks hundreds of technology identifiers back to 2010 plus shared hosts, which surfaces a company's other brands and its agency), the subdomain roster, trends pages with live against expired counts, provenance (BuiltWith says it indexes the internet the way a search engine does, refreshing weekly, and takes no third-party data, so a domain blocking indexing bots is tagged rather than silently thinned), and the vendor's own published one-off report route. Legal section covers the two places this platform stops being research: the meta export's named-individual columns, and uploading your own customer list to a third party.
- [Using Wappalyzer for Competitive Intelligence](https://www.flares.tech/sources/wappalyzer): Separates the two products that share the name, because most reported disagreements are somebody comparing one against the other. The browser extension analyses websites locally in your browser (Wappalyzer's own privacy policy), so it reads the response you personally received, including pages behind a login, a trial account or a regional variant that no external index reaches. The shared index is a different instrument and is filtered before you see it: the published API documentation shows low-confidence detections excluded by default, results drawn from what was verified in roughly the last two months by default, lookups following internal links so a site answer spans several pages, and cached results returned rather than live scans, with the vendor describing its cache as both faster and MORE COMPLETE than a real-time read. Consequence stated as an operating rule: never prove a negative at default settings, because two of those four filters suppress results by design. Documents where the index comes from, which nobody writes: with sharing enabled the extension periodically sends hostname, protocol, detected technologies and detection metadata, and the policy states that data may be shared or sold, which explains why the free technology pages list logged-in application hostnames and internal consoles no link-following index would reach, and why an analyst researching with sharing on contributes their own employer's internal hostnames. Also records, from three checkable public sources, that the detection ruleset went private in August 2023: the original repository no longer resolves, the public package's final version is dated 23 August 2023 and carries a no-longer-supported notice, and a third-party copy described as the last commit before it went private was created the same day under the original licence, with a separately maintained fork still current. Consequence for method: a label can no longer be audited against its signature, so confirm it against the page.

## Free competitive intelligence tools

- [Feature Parity Comparison Matrix Template](https://www.flares.tech/competitive-intelligence-tools/feature-parity-comparison-matrix-template): Build a feature comparison matrix across competitors.
- [Competitive Analysis Checklist](https://www.flares.tech/competitive-intelligence-tools/competitive-analysis-checklist): A step-by-step checklist for running a thorough competitive analysis.
- [Win/Loss Analysis Questions](https://www.flares.tech/competitive-intelligence-tools/competitive-win-loss-analysis-questions): A curated set of win/loss interview questions to understand why deals are won or lost.
- [Competitive Intelligence Audit Scorecard](https://www.flares.tech/competitive-intelligence-tools/competitive-intelligence-audit-scorecard): Score the maturity of your competitive intelligence program.
- [Positioning Map Template](https://www.flares.tech/competitive-intelligence-tools/positioning-map-template): Plot competitors on a two-axis positioning map.
- [SWOT Analysis Template](https://www.flares.tech/competitive-intelligence-tools/swot-analysis-template): Interactive builder for strengths, weaknesses, opportunities, and threats, with PNG/PDF export.
- [Competitive Pricing Matrix Template](https://www.flares.tech/competitive-intelligence-tools/competitive-pricing-matrix-template): Compare competitor pricing tiers side by side.
- [ICP Template](https://www.flares.tech/competitive-intelligence-tools/icp-template): Define your Ideal Customer Profile.

## Optional

- [Early Access](https://www.flares.tech/early-access): Request early access to Flares.