How to Use CRM Data for Competitive Intelligence

A CRM is the only system in the company that knows the outcome of every deal you contested, how long each took, what you charged, what you conceded and who else the buyer was talking to. Described that way it sounds like the richest competitive source available, and in principle it is. It is also the source that most reliably disappoints, because the last item in that list is usually missing.

The distinction worth holding on to is between what a CRM records automatically and what somebody has to decide to record. Stage history, activity, quote versions, closed values and discounting all accumulate whether anybody thinks about competitors or not, which makes them trustworthy. The competitor name, the loss reason and the useful half of the deal notes are all typed in by a person at the end of a process that did not go their way, which makes them patchy in a direction that is not random.

So the honest description of this source is that it is excellent evidence about your own commercial behaviour, moderate evidence about which rivals you meet, and poor evidence about why anything happened. Used in that order it will produce numbers nobody can argue with. Used the other way round, starting from the reason codes, it produces a confident account of your market assembled entirely from the beliefs of the people who lost.

1. 1Check whether a competitor field exists before analysing anything. Open a closed-lost record and look. In many systems there is a reason for the loss and nothing that names who won, which means every competitive question you were about to ask has no data behind it. Establish this first, because it changes the project from analysis to instrumentation.
2. 2Make the competitor a picklist, separate from the loss reason. Two fields, not one. A single list mixing 'price' with 'lost to a named rival' cannot answer either question, and free text cannot be counted at all. A short controlled list of named competitors plus 'no decision', 'internal build' and 'other' covers nearly every deal and stays countable.
3. 3Require it at the stage where the answer is known. Requiring the field at deal creation captures a guess, because nobody knows the shortlist yet. Requiring it at the close stage captures an answer, and it is the one stage where a mandatory field is tolerated because the deal is already over. Anything optional is filled in on roughly the deals where it least matters.
4. 4Backfill from the sources that already hold the answer. Do not start the history from today. Deal notes and recorded calls usually name the rival on deals where the field is empty, and one focused pass over the last four quarters of significant losses will produce a usable baseline. Mark backfilled records so nobody later mistakes reconstruction for what was recorded at the time.
5. 5Produce four numbers per named competitor, with their denominators. Win rate against them, median cycle length when they are present, median discount when they are present, and average deal size in those deals. Each one is a comparison against your own baseline rather than a figure in isolation, and each must be published with the number of deals behind it or it will be quoted long after the sample stops supporting it.
6. 6Split no-decision out of competitive losses. Deals where nobody bought anything are a different phenomenon with a different fix, and folding them into competitive losses inflates your apparent loss rate to every rival at once. Report them as their own category and watch the share: it is frequently the largest single outcome in a B2B pipeline.
7. 7Validate the reason codes against somebody who was not the seller. Take the loss reasons for one quarter and compare them against buyer interviews on the same deals. If the two disagree, and they usually do, the codes are recording what your sellers believe rather than what happened. That is still worth knowing, provided you label it as such rather than reporting it as the cause.

Before blaming a sales team for poor data hygiene, it is worth checking what the software actually asked them to fill in. HubSpot’s own documentation lists Closed lost reason among the default deal properties, described as the reason the deal was lost. There is no default property that names the competitor who won it. Salesforce is the more interesting case: it ships a standard object, OpportunityCompetitor, for recording competitors against an opportunity, and publishes training material on competitor tracking. The related list still has to be added to the page layout and populated by hand.

Both vendors, in other words, treat the competitive layer as something a company opts into. That single design decision explains most of what people describe as a cultural problem. A seller closing a deal is shown a form. If the form has a required field naming the rival, it gets filled in on nearly every deal. If it has no such field, the information goes into a free-text note if it goes anywhere, and no amount of asking people to be more diligent will produce a countable dataset from that.

The two-field rule

The most common way this gets implemented badly is with one field doing two jobs. A loss reason list that mixes “price” and “budget cut” with the names of three competitors cannot answer either question: you cannot count how often you lose to a named rival, because some of those deals were coded “price”, and you cannot count price losses, because some of them were coded with a competitor’s name. Two fields, one naming who won and one giving the stated cause, is the whole fix, and it takes minutes.

Once the field exists, the analysis is short. Four measures, each computed as a comparison against your own baseline rather than as a figure standing alone, and each published with the number of deals behind it. Everything else this source can do is a refinement of these.

1. Win rate against them

The obvious one, and the one that needs the most caution. Use closed deals only, exclude outcomes where nobody bought, and quote the count next to the percentage every single time. Below roughly thirty closed deals against one rival, a single deal changing hands moves the figure several points, which is how a slide ends up reporting a trend that is really a rounding artefact.

2. Median cycle length when they are present

Recorded automatically, so it does not depend on anybody remembering anything, and it is a good early indicator because it moves before win rate does. A rival who adds three weeks to your median cycle is imposing a cost on every contested deal whether you win or lose, and that cost is invisible in any measure that only counts outcomes.

3. Median discount conceded when they are present

The most useful number in the whole system, and the one almost nobody computes. Take the median discount on deals where a named competitor appeared and subtract the median on deals where none did. The difference is what that rival costs you per deal, in your own currency, measured rather than estimated, with no assumption about their price list required.

4. Average deal size in contested deals

Read alongside the discount figure rather than on its own, since the two move together. If contested deals are both larger and more discounted, a rival is pulling you into a segment where you compete on price; if they are smaller and equally discounted, you are being used to set a floor. Working out what the competitor charges is a different exercise, covered in competitor average deal size.

The comparison table grades all eleven. What follows is what each one is for, including the four that carry most of the value and are almost never queried.

1. The competitor field on an opportunity

Everything else on this page depends on it existing and being populated at close. Keep the list short, controlled and inclusive of the non-vendor outcomes: “no decision”, “internal build” and “other”. Review what lands in “other” each quarter, because a rival appearing there repeatedly is a new entrant your list has not caught up with.

2. The closed-lost reason

A record of your sellers’ beliefs. That is genuinely useful, provided it is labelled honestly: knowing that your team thinks it loses on price tells you what they will ask for next quarter, whether or not it is true. As a picklist it can be counted; as free text it is readable one deal at a time and useless in aggregate.

3. Free-text deal notes and call summaries

Where the competitive detail actually lives on deals closed before anybody added a field. Search them for competitor names during a backfill and you will usually recover most of a year’s history. The notes also carry things structured fields never will: what the rival offered, when they entered, and what the buyer said about them. The fuller version of that material sits in sales call recordings.

4. Stage history and time in stage

Recorded by the system rather than by a person, which makes it the most trustworthy competitive measure available here. Beyond total cycle length, look at which stage stretches: a deal that sits in evaluation twice as long against one rival is telling you where their product comparison hurts.

5. Discount and approved-price fields

Covered above as the fourth number. One practical note: approval records are often better populated than discount fields, because approvals have a workflow attached and somebody had to click. If the discount field is patchy, the approval log will frequently reconstruct it.

6. The quote or proposal record

Version history is the useful part. The point at which a quote was revised, and by how much, dates the moment a competitor forced a commercial change. Across a set of deals against the same rival, those revisions cluster at a particular stage, which tells you when in their process they make their pricing move.

7. Products and quantities on closed deals

Where your range wins and where it does not, by rival. The most valuable pattern to look for is a turnover point: the seat count, volume band or module combination above which your win rate against a specific competitor drops sharply. That threshold is a product and pricing finding at once, and it is invisible at the level of an overall win rate.

8. Renewals, downgrades and cancellations

A competitor taking an existing customer is a different and more expensive event than one winning a new deal, and it is recorded in a different object that competitive reporting usually ignores. Read the elapsed time between the first competitive mention in the account and the cancellation, since that window is what a retention response has to fit inside. The pattern across accounts is what competitor churn work is built from.

9. Reopened and won-back opportunities

The most under-used records in the system. A deal a rival won and subsequently lost back to you is direct evidence of where their product or service fails after the sale, from a buyer who experienced both. There are never many of these, and each one is worth reading in full rather than counting.

10. Lead source and campaign attribution

Graded low because attribution is contested in every company that has it, but useful for one narrow question: which acquisition channels bring deals that a named rival also contests. That tells you where you and they are genuinely meeting, as opposed to where you assume you compete.

11. The activity log

Meeting and email counts per deal, and more usefully the gaps. A long silence before a loss dates the moment the buyer stopped considering you, which is frequently weeks before the deal was formally closed and well before anybody internally noticed. Comparing that silence point across deals against one competitor shows you where in your process you lose them.

What it costs

Nothing in licence terms, since you already own the system. The cost is setup and maintenance: an hour to create two fields and a required-at-close rule, two to four days for a proper backfill of the last four quarters, and about half a day each quarter to produce and publish the numbers. The backfill is the part that gets cut, and cutting it means waiting a year before any trend exists.

Permissions

Reporting on closed opportunities usually needs less access than people assume, and a read-only reporting role over closed deals is normally enough. Ask for that rather than for broad record access: it is easier to approve, and it keeps you out of live deals, which is where sales leadership’s objections actually come from.

Reports rather than downloads

Build the four numbers as saved reports inside the CRM so they refresh, and take a spreadsheet copy only when you need to join deal records to something the system does not hold, such as interview findings or a rival’s published prices. A one-off download that becomes the number everyone quotes is stale within a quarter and nobody remembers which filters produced it.

Which systems support it

Salesforce has the standard competitor object waiting to be enabled. HubSpot needs one custom property. Pipedrive and Attio both support custom fields and filtered reports that do the same job. The software is almost never the constraint; the required-at-close rule and somebody publishing the output are.

The records are yours, so the usual questions about how material was obtained do not arise here. What does arise is that a CRM is unusually widely readable inside a company, retained for years, and routinely produced in litigation and due diligence. Anything written into it is written into a document with a long life and a broad audience, and that is the test to apply to every competitive note.

• Never record a competitor’s confidential information. If a buyer forwards you a rival’s proposal marked confidential, the deal record is the worst possible place for it. Note that a competing proposal existed and what shape it took; do not attach the document and do not transcribe its terms.
• Nothing about pricing intentions exchanged with a competitor. A note describing what a rival told you directly about their future pricing turns an ordinary CRM record into evidence of contact that competition authorities take seriously. Information reaching you through the market is fine; information reaching you from them is a different category and does not belong in your records or your process.
• Keep speculation about named individuals out. Notes characterising a competitor’s employees are personal data, contribute almost nothing to a competitive picture, and read appallingly when produced years later. Record the role and the behaviour, not the person and your opinion of them.
• Treat the buyer’s data with the same care as the deal data. Contact records carry personal data with purpose limitation and retention obligations attached. Competitive analysis is a legitimate internal use of the deal record; it is not a reason to keep a contact’s details indefinitely after the relationship ends.
• Write every note as if it will be read aloud. This single habit resolves most of the above without a policy document. Sellers write faster and more candidly than anybody expects, and a competitive programme that encourages colourful notes about rivals is creating a liability while it collects data.

• The market you never reached. Your records contain deals you were in. A rival winning steadily in a segment where nobody invites you produces no rows at all, and the absence looks identical to strength. Proxy: their published customers and the segments their commercial hiring targets.
• Why anybody decided anything. The only causal field is filled in by the person who lost, from the outside of a decision they were not present for. Proxy: an interview with the buyer, conducted by somebody who was not the seller.
• Anything about the competitor as a business. Their revenue, headcount, funding and roadmap leave no trace in your pipeline, however complete it is. Proxy: filings, registries and their own published material.
• Deals that never became opportunities. Buyers who evaluated the category and chose a rival without ever contacting you are absent, and in self-serve segments they outnumber the ones who did. Proxy: what the category is searched for, and the comparisons people run in public forums.
• History, if the field is new. A field created this quarter has no past, and a trend needs at least four. Proxy: a backfill from deal notes and recorded calls, clearly marked as reconstructed.

Publish the four numbers every quarter, with their denominators, to the people who fill the field in. That is the whole maintenance mechanism, and nothing else substitutes for it: a field whose output is never shown degrades within two quarters regardless of how the validation is configured. Sellers keep populating a field when they have seen a pricing rule or an enablement asset change because of it.

Add one audit pass per quarter over the deals where the competitor is blank or set to “other”, since those two buckets are where the quality problem always hides, and a name recurring in “other” is a new entrant your picklist has not caught up with. Keep a dated snapshot of each quarter’s figures outside the CRM as well, in a competitor tracking spreadsheet or equivalent, because CRM reports get edited and a comparison you cannot reproduce is a comparison nobody will trust.

Every field discussed on this page is written at the end. The competitor is recorded when the deal closes, the discount when it is approved, the reason when the outcome is already fixed. That makes the CRM an excellent historian and a hopeless early-warning system. It will tell you precisely who beat you last quarter and it cannot tell you that a rival changed their entry price three weeks ago, which is the fact that decides the deals currently open. By the time a competitive shift is visible in your pipeline reporting, you have already paid for it in deals nobody knew were being contested on new terms.

To see it earlier you have to watch the rival instead of the record of your own deals, and that is a monitoring job rather than a reporting one. It is the reason to run competitive intelligence software alongside the pipeline at all. Flares surfaces what each rival changes on price, product and positioning as it happens, so a shift reaches a seller while their deal is still open rather than appearing in a quarterly review. No outside system knows what those deals were worth to you, what you conceded to hold them, or how long they took. Those figures live in one place only, and only if somebody made the field required.