How to Analyze Competitor Strategy: 10 Sources and Four Corners

Every guide to this question opens the same way: identify your competitors, audit their products, review their marketing, draw conclusions. The steps are not wrong and they produce a description of the present, which is not what anybody asked for. The question behind this search is almost always about the future, and specifically about one company’s future behaviour towards you.

Two things change the quality of the answer. The first is that strategy is frequently published rather than inferred: a listed company writes down its growth strategy, names its competitors and lists what it believes the market buys on, all under legal liability, in a document you can read for nothing. The second is that where nothing is published, the reliable evidence is not what a company says but what it has already paid for, because commitments carry information that statements do not.

1. 1Write the question as a decision, not as a topic. Analyse their strategy is not a question. Will they follow us into the mid-market, will they cut price if we launch a free tier, can they ship this in a year: those are answerable and they tell you which evidence to go and get.
2. 2Take their own words first, if they publish any. For a listed competitor the annual report contains a strategy section, a competition section and a risk section, all written by them and all reviewed by lawyers. Reading those three takes an hour and skips a fortnight of inference.
3. 3Fill in all four corners, not just the visible one. Future goals, current strategy, assumptions and capabilities. Most analyses describe the current strategy and stop, which produces a summary of the present rather than any view of what happens next.
4. 4Rank every piece of evidence by how hard it is to reverse. A blog post can be deleted in an hour. A price change takes a quarter to undo. A senior hire takes a year. An acquisition never reverses. When evidence conflicts, the least reversible commitment is the one telling the truth.
5. 5Diff the same document across two consecutive years. Put last year's competition and strategy sections beside this year's and read only what changed. Additions to a list of competitive factors are a company stating, in writing, what it now believes the market buys on.
6. 6Write the response profile, not the report. For each move you might make, state what they will probably do, within what horizon, and which of the four corners supports it. This converts a description into something a product or pricing decision can actually use.
7. 7Date it and record what would prove you wrong. Every prediction gets a horizon and a disconfirming signal written beside it. Then check. A strategy analysis nobody ever scores is an opinion that accumulates confidence without ever being tested.

Strategy is not a data point you go and find. It is the pattern that emerges when ten separate observable layers are put beside each other and read in one direction. Doing it layer by layer also stops the usual failure, where a team studies whichever layer is easiest to see and calls the result a strategy.

You do not need all ten to say something useful. Three or four layers pointing the same way is a finding, and one layer moving against the others is usually the most interesting thing on the page: a company whose pricing says one thing and whose hiring says another is mid-decision, and that is the moment when your own move has the most leverage.

Michael Porter’s four corners model is the framework built specifically for this question, and it is worth using for one reason: its output is a prediction of competitor response rather than a profile. Two of the corners describe what a company does, and two describe why, which is why analyses that skip the second pair keep being surprised.

For a listed competitor, stop inferring and start reading. The annual report contains a business section that usually includes an explicit growth strategy, a competition section describing the market and naming rivals, and a risk section listing what management believes could go wrong. All three are written by the company, reviewed by its lawyers, and updated every year. This is not a rare disclosure: the exact phrase “our principal competitors” appears in over 8,500 annual reports in the regulator’s full-text index, across every industry.

The technique that turns this from reading into analysis is the diff. Open the same section in two consecutive filings and read only what changed, because the additions are a company stating in writing what it now believes it is competing on.

Three things follow from a diff like that, and none of them required an interpretation. The company has told you what it now thinks buyers evaluate on. It has told you which new kind of company it considers a competitor, which is a statement about the shape of the market rather than about any one rival. And by putting those items in a legally reviewed document, it has committed to a story it will have to keep telling. Do the same exercise on the risk factors and you get the assumptions corner filled in for free.

The private-company version of this exists in more places than people expect. In the United Kingdom, companies that are not entitled to the small companies exemption must file a strategic report with their accounts, containing a fair review of the business and a description of the principal risks and uncertainties. That is a filed, dated strategy statement from a company that has never issued a press release about its plans.

1. The strategy section of their annual report

Usually titled growth strategy or business strategy, and usually a list of four to six numbered intentions with a paragraph each. Read it for what is first and what is missing, since order in these lists is rarely accidental and an intention that quietly drops out between two years is a retreat nobody announced. Take the numbers embedded in it too, because companies illustrate strategy with disclosures they do not repeat elsewhere, such as the share of customers outside their home market and the share of revenue those customers generate.

2. The competition section of the same filing

The most under-read page in competitive intelligence. It names the categories of company they consider rivals, frequently names specific firms, and lists the factors they believe the market competes on. That last list is their view of the buying decision, stated by them, which is a direct answer to the assumptions corner. Read it beside your own view of the buying decision and the differences are where one of you is wrong.

3. Risk factors and principal risks

Long, defensive and full of boilerplate, and still worth the hour. The technique is to ignore everything generic and look for risks written in unusual specificity, because those are the ones drafted about a real situation rather than copied from a template. A risk factor that names a technology, a customer concentration or a particular regulatory outcome tells you what management was arguing about when the document was written. Compare against last year and note what was added.

4. Investor days, shareholder letters and earnings calls

Where strategy is described with numbers attached and a horizon stated. Investor day decks carry multi-year targets by segment, which is the closest thing to a published plan you will ever get. Earnings calls add the unscripted half: analysts ask what management is avoiding, and the answers, including the evasive ones, are informative. Search transcripts for the same phrase across four quarters and watch how the framing shifts, which is often visible before the strategy is.

5. National company registers and filed accounts

The route into private competitors, and specific to jurisdiction. Larger UK companies file a strategic report with a fair review of the business and its principal risks. German and French registers carry accounts and officer records. What you are looking for is not the financial detail but the narrative sections and the structural changes: a new subsidiary in a country, a change of registered activity, a capital increase timed alongside a product launch.

6. Their pricing, packaging and product surfaces

Strategy as executed. A new tier, a repackaged bundle, a feature moved between plans and a changed minimum commitment are all decisions that took a quarter of internal argument, and each is visible the day it ships. This layer is the fastest of them all and it is the one to check first when something else suggests a change, because pricing is where a strategic decision becomes real for customers. It is also where a claimed strategy and an actual one most often diverge.

7. Job adverts and leadership appointments

The best forward-looking evidence available for free. A function being staffed for the first time tells you it is now resourced. Several roles opened in the same area over one quarter means a budget was approved rather than a view was expressed, and a first appointment at executive level is the strongest version of both. Read the seniority mix as well as the count, since ten junior roles and one vice president mean different things. The reading method for individual appointments sits under competitor leadership team.

8. Patents, trademarks and other paid-for filings

Slow, lagging and unusually honest, because each one costs money and professional time. A cluster of filings in one technical area is a bet somebody signed off. Certification and authorisation programmes belong in the same category and are faster to read: a competitor appearing on a government cloud authorisation list has committed to a market months before any revenue arrives, as covered under vendor security certifications. Absence proves nothing in either case, so use these to confirm a direction rather than to discover one.

9. Executive interviews, podcasts and conference talks

An hour of unscripted talking produces material no document will. Executives explain their reasoning, describe what they tried and abandoned, and say which competitor they take seriously. Search the person’s name rather than the company name, and prefer long formats, since brevity is where messaging discipline survives. Treat everything said here as intent rather than commitment, and check whether it later appears in a filing or in the product.

10. Your own win/loss record and deal notes

The only source that tests your analysis instead of feeding it. Every deal where you met them is an experiment: what they discounted, which objection they led with, whether they brought in an executive, how quickly they responded. Over two quarters this tells you how they behave under pressure, which is exactly what a competitive response prediction is about. If your file says they never discount and your sellers say otherwise, your file is the thing that is wrong.

The deliverable is not a document about them. It is a short table about you, listing the moves you are considering and what each one is likely to provoke. That reframing is what gets the work read, because a product or pricing decision can act on it directly.

Rank the evidence by how hard it is to reverse

Sources contradict each other constantly, and you need a rule for which one wins that is not simply the most recent. Reversibility works better than anything else: how much would it cost this company to undo the thing you are looking at. A blog post can be deleted before lunch and means almost nothing. A pricing change takes a quarter to unwind and embarrasses somebody. A senior hire takes a year and a severance conversation. A new office, an acquisition or a certification programme is effectively permanent on any timescale that matters to you.

So when the messaging says up-market and the hiring says down-market, believe the hiring. When the conference keynote says platform and the pricing page still charges per seat for one product, believe the pricing page. This single rule resolves most of the contradictions a real analysis produces, and it keeps you from being moved by announcements, which are the cheapest signal any company can emit.

1. 1Separate observation from inference, in the file. Two fields, always. What was observed, with a source and a date. What you concluded, with a confidence level. Merging them is how a guess becomes a fact over three quarters.
2. 2Find the second layer that agrees. One signal is an anecdote. A pricing change plus matching hiring plus a changed risk factor is a direction. Require two independent layers before anything reaches a slide.
3. 3Write the disconfirming signal down. What would you expect to see within two quarters if you are wrong. Naming it in advance is the only protection against reading every subsequent event as confirmation.
4. 4Argue the opposite case out loud. Have somebody who did not do the research make the strongest argument that the competitor is doing the reverse. Half an hour of this catches more errors than another week of collection.
5. 5Score your last prediction first. Before writing a new analysis, go back and mark the previous one right or wrong. Teams that skip this produce steadily more confident and steadily less accurate work.

Reading published material and reasoning about it triggers nothing anywhere. The unusual feature here is that the exposure sits in the document you produce rather than in the material you collect. A strategy analysis is speculative by nature, it names a competitor, and it is written in a file that can be read by people you did not intend.

• Write as though it will be read aloud. Internal documents are routinely produced in litigation and in due diligence. A line describing a rival’s conduct in loaded language, or a note about exploiting a weakness that reads badly out of context, is a liability with no analytical value. State facts, sources and dates, and keep the characterisation out.
• Label inference as inference. A prediction recorded as a fact will eventually be repeated to a customer as a fact, and a false statement about a competitor’s plans is a different kind of problem from an honest analysis that turned out wrong. Two fields, one for what you saw and one for what you think.
• How you obtained it matters more than what it says. Public information assembled into a conclusion is lawful everywhere. Information obtained through deception, payment, or by persuading somebody to break a confidentiality obligation is misappropriation regardless of how ordinary the fact turns out to be.
• Be careful around listed competitors. If research surfaces material information about a public company that is not yet public, what you and your colleagues do next is a securities question rather than a competitive one. Stop, record how it reached you, and take advice before it circulates further.

• The actual plan document. It exists, it is a deck, and it is not obtainable through any legitimate route. Proxy: the published strategy section plus investor day targets, which are the sanitised version of the same content.
• Timing. Direction is readable and dates are not, because internal dates move constantly. Proxy: hiring, which converts into shipped work on a fairly consistent two to four quarter lag.
• What they decided not to do. Rejected options are the most useful part of any strategy and they are never disclosed. Proxy: capabilities they built and then stopped investing in, visible as a product area with no releases and no hiring.
• Internal disagreement. Whether the leadership team is aligned changes everything about how fast they can move, and it is invisible. Proxy: executive departures clustered in one function, and public messaging that contradicts itself between executives.
• Their view of you. Whether you are a named threat internally or an afterthought is not published. Proxy: whether they build comparison pages against you, bid on your brand terms, or send an executive into deals where you appear.

Run three clocks. Annually, rebuild the four corners from scratch rather than editing last year’s, and keep the old version, because the difference between two years of your own analysis frequently says more than either version does. Quarterly, refresh the fast layers: pricing, packaging, hiring and any new filing. Continuously, add observations to the file with dates, since the value of this work compounds only if the record is unbroken.

Five events justify reopening the whole analysis immediately: a funding round, an acquisition on either side, a change of chief executive or revenue leader, a repricing, and a public entry into a market they were not in. Each of these resets at least one corner, and an analysis carrying a stale corner is worse than no analysis, because it is quoted with the same confidence as a current one.

Strategy does not decay the way a price does. It decays by accumulation: a packaging change here, a run of job adverts there, a sentence added to a risk factor, an executive using a word they did not use last year. No single one of those is worth an alert, and eight of them across two quarters are a company changing direction. The failure mode is unusually specific and unusually expensive, because nothing looks wrong at any point. The analysis in your file stays fluent, confident and quietly detached from the company it describes, and you find out at the exact moment a customer tells you what your competitor is now doing.

Holding ten layers across several companies over years is a memory problem rather than an analysis problem, and it is the reason strategic intelligence tends to survive only where something keeps the record. Flares holds competitor pricing, product, messaging and hiring changes in one unbroken timeline, so the drift shows up as a sequence you can read rather than as a gap you have to reconstruct. What it will not do is the corner that matters most. Working out what a competitor believes, and where that belief is wrong, is a judgement somebody has to make and defend, and it is the part of this work worth protecting time for.