A CEO I work with asked his VP of Sales a simple question in a board prep session: "Are we losing more to Competitor A or Competitor B?"
The VP said A. The head of marketing said B, because that is who kept showing up in demo requests. Somebody pulled up HubSpot and filtered closed-lost deals by the competitor field. Of 214 lost deals in the last four quarters, 38 had a competitor named. Eighteen percent. The rest were blank, or said "other", or had a free-text entry like "the incumbent tool".
So the honest answer to the board was: we do not know. Not "we lose 60% to A". Not "B is the real threat". We do not know, and we have not known for two years, and every strategy conversation about positioning has been running on vibes.
That is the actual state of competitive intelligence at most B2B companies under 200 people. Not a missing battlecard. Not a missing tool subscription. A missing data layer. Crayon's State of Competitive Intelligence research found that 44% of companies have no competitor visibility inside their CRM at all, which means nearly half of B2B revenue teams are in exactly the room I just described.
Of companies have no competitor visibility inside their CRM. They can tell you their overall win rate to two decimal places and cannot tell you who they lose to.
This post is about building the part everyone skips. Not what a competitive intelligence program should contain in theory, but what to instrument first so the program has something real to run on.
Why most CI programs start in the wrong place
The standard sequence goes like this. Somebody loses a painful deal. Leadership asks for "better competitive enablement". Product marketing spends six weeks building competitor pages in Notion. There is a launch in the sales channel. Three months later nobody has opened them, the win rate has not moved, and the conclusion is that the team needs a proper CI tool.
The tool gets bought. It watches competitor websites, pings Slack when a pricing page changes, and generates cards. Twelve months and $20,000 later the same CEO asks the same board question and still cannot answer it, because the tool watches the outside world and the answer lives in the inside world. Who you actually competed against, in which deals, at what size, and what happened. That data was never captured, and no vendor can retroactively create it for you.
Content is the visible part of competitive intelligence. Data is the part that decides whether the content is aimed at anything.
A competitive intelligence program is a measurement system with content attached, not content with measurement attached.
If you cannot compute win rate against a named competitor by segment, everything downstream is guesswork with a nice template. Build the capture layer first. It costs a week and it changes what you argue about in every pipeline review after that.
The capture layer, in detail
This is the unglamorous work. It is also the only part that compounds, because every month of clean capture makes the next analysis better and no amount of budget buys back the months you did not capture.
Put competitor data on the deal, not the company
The most common modelling mistake I see is a competitor property on the company or account record. It feels tidy and it destroys the analysis. Acme might evaluate you against one competitor in their first purchase and a completely different one when they expand eighteen months later. Competition is a property of a specific buying decision, so it belongs on the deal.
In HubSpot, create these on the deal object:
- Primary competitor. Single-select dropdown, not free text. Free text gives you "Salesforce", "SFDC", "salesforce.com" and "sf" as four different competitors within a month.
- Other competitors considered. Multi-select, same option list. Most real evaluations have two or three names in them, and the second name is often the one quietly winning.
- Competitive status. Won against, lost to, still competing, no competitor, unknown. This is the field that separates a genuine head-to-head loss from a no-decision, which matters more than people expect.
- Competitor context. A short text field for the one sentence that explains the deal. "They already used their reporting module" is worth more than any dropdown.
Keep the dropdown list short. Five to eight named competitors, plus "in-house or spreadsheets", plus "no decision", plus "other". If you list 30 competitors, reps pick "other" because scanning the list costs more than the field is worth to them. Review the list quarterly and promote whatever keeps showing up in the context field.
Decide when the field is required
Requiring a competitor on every new deal is how you get garbage. At first contact the rep genuinely does not know, so they pick something to get past the validation, and now your data is confidently wrong, which is worse than blank.
Gate it at the stage where the answer exists. For most B2B teams that is proposal or later. The rep cannot move a deal into proposal without setting competitive status, and if the status is "still competing" or "lost to", primary competitor becomes required. Deals that close before proposal keep an honest "unknown".
This single rule took one client from 18% competitor coverage to 81% in a quarter, with no enablement campaign at all. Reps do not resist data entry. They resist data entry that asks them for something they do not have yet.
Stop relying on rep memory as the only source
Even with good field discipline, self-reported competitor data has a bias. Reps under-report competitors in deals they lost badly, because naming a competitor invites a conversation about the deal. They over-report price-based competitors because "they went cheaper" is a comfortable story.
If you already run conversation intelligence, you have a second, unbiased source sitting in your call recordings. Gong, Clari Copilot and Fathom all support competitor keyword tracking. Set up trackers for every name on your dropdown list plus their common misspellings, then run a monthly reconciliation: deals where a competitor was mentioned on a call but the deal record says "no competitor". That list is your data quality report, and it also tends to surface competitors you did not know you had.
The four numbers you can finally compute
Once capture is running, you get access to a set of numbers that most companies your size simply do not have. Each one answers a question that leadership is already arguing about.
Competitive win rate by competitor. Closed-won divided by closed-won plus closed-lost, filtered to deals where that competitor was present. Exclude no-decisions and unopposed wins, and report those separately. An overall win rate of 24% can hide 55% against one competitor and 9% against another, and those two numbers imply completely different strategies. Our guide on how win rate gets inflated covers the counting traps here.
Competitive win rate by segment. This is where it gets useful. Slice by deal size band, industry, and inbound versus outbound source. The pattern I see most often: you beat the enterprise competitor comfortably under 40 seats and get destroyed above 150, or you lose to the cheap tool in inbound and never see them in outbound. Both findings change your targeting, not your messaging.
Cycle length delta. Average days to close in competitive deals versus uncontested ones. If competitive deals take 40% longer, that is real capacity being consumed, and it is an argument for either better qualification or better early positioning.
Discount delta. Average discount given in deals against Competitor A versus your baseline. I have seen teams discover they were giving eleven points of margin away against a competitor they were beating 70% of the time. That is money handed over out of habit rather than necessity.
Build these as one dashboard, not four reports. If you use HubSpot, a custom report on deals grouped by primary competitor with a closed-won rate calculation gets you most of the way. The point is that this dashboard becomes the input to your quarterly strategy discussion instead of whoever spoke most confidently.
The monitoring layer, and how much to spend on it
Now the outside-world part, which is what people usually mean by competitive intelligence. It matters, but it is second, and it is cheaper than the vendor market implies.
For a team under 100 people, most of the signal comes from five sources you can watch with automation you already own:
The first four are a weekend build in n8n. Scheduled HTTP requests to the pages you care about, a hash comparison against last week's version, an LLM node to summarise what changed in two sentences, and a post into a #competitors Slack channel with the diff attached. Cost is close to nothing beyond a few cents of model tokens and whatever your n8n instance already runs on. We build this pattern for clients as part of AI and automation work and it consistently outperforms the alert quality of paid tools, because it watches ten pages you chose rather than a thousand you did not.
The fifth source is win/loss interviews, and no software replaces it. Ten interviews a quarter with buyers who chose someone else will teach you more about your competitive position than any monitoring feed. Have someone who is not the deal owner make the calls.
When a CI platform is actually worth it
Klue, Crayon and Kompyte are real products and I am not against buying them. Klue entry deployments generally start around $15,000 to $20,000 a year and go up sharply, Crayon sits in a similar band, and Kompyte comes in materially cheaper. The question is whether you have the problem those price tags solve.
Buy one when you have a named person who owns competitive intelligence at least half time, more than five competitors that genuinely move revenue, and reps who are already using competitor data in deals. The platform then saves that person real hours and pushes cards into the seller workflow properly.
Do not buy one when nobody owns it. A CI platform with no owner produces a Slack channel of alerts that everyone mutes in the second week. I have watched three companies do this. In every case, the money would have paid for a part-time product marketer and the outcome would have been better.
Turning intel into something reps use
Once the data is real, the content problem gets much easier, because you know which two competitors deserve the effort. Everything I would say about the format is in the battlecards guide, so here is only the part that connects to the data layer.
Trigger the intel off the field you just built. In HubSpot, a workflow on primary competitor can post the relevant card link into the deal record and notify the rep, so the card arrives when the competitor is named rather than during onboarding six months earlier. That is the whole trick. The card does not need to be better. It needs to arrive at the moment the field gets set.
Then close the loop. Every quarter, take the deals where a card was triggered and compare win rate against deals with the same competitor where it was not. If the difference is nothing, the card is wrong and you now know that from evidence instead of opinion.
A 30-day rollout that actually finishes
Week one: model the fields. Deal properties, short dropdown, stage gate at proposal. Get the sales leader to approve the competitor list in a fifteen-minute conversation, not a workshop.
Week two: backfill what you can. Pull the last two quarters of closed-lost deals and have reps tag competitors in a single 45-minute working session. Bulk edit, not one at a time. Partial backfill beats waiting a quarter for fresh data, though flag it as lower quality when you report on it.
Week three: build the dashboard and the call trackers. Four numbers, one screen. Set the trackers in your conversation tool for every name on the list.
Week four: build the monitoring workflow in n8n for your top three competitors only, and write cards for the two competitors the data says matter. Ship those two and ignore the rest until the numbers say otherwise.
You will not have a mature program in 30 days. You will have something better than most companies your size have after two years, which is a competitor number you can defend in a board meeting.
Mistakes worth avoiding
Tracking too many competitors. Five that move revenue beats twenty that appear occasionally. The long tail eats time and produces nothing.
Treating the competitor field as a reporting requirement. If reps get nothing back, coverage decays within a quarter. Show them the win rate against their own competitors in the pipeline review. Data that comes back to you is data you keep entering.
Letting product marketing own the number. They should own positioning. RevOps should own the field design, the validation rules, and the dashboard, because the same CRM data quality discipline applies here as everywhere else.
Confusing competitor monitoring with competitive strategy. Knowing that Competitor A changed their pricing page is not a decision. Knowing you lose to them 78% of the time above 100 seats is.
Building it once. The competitor list at a growing company is different every year. If you have not changed a dropdown option in eighteen months, you are tracking last year's market.
Cannot answer the board's competitor question either?
We instrument competitor capture in HubSpot or Salesforce, backfill what your history supports, and hand you the four-number dashboard. Usually two weeks, fixed scope.
Book a RevOps audit →Frequently asked questions
What is competitive intelligence in B2B sales?
It is the practice of collecting and using information about the competitors your buyers actually evaluate, so you can position against them and decide where to compete. In practice it has three parts: internal data about who you meet in deals and how those deals end, external monitoring of what competitors are doing, and enablement content that gets that knowledge to reps during live calls. Most teams build the third part first and wonder why it does not work.
How do I calculate competitive win rate?
Take closed-won deals where a specific competitor was recorded, divide by the total closed-won plus closed-lost deals with that competitor, and exclude no-decision outcomes and deals where you had no competition. Report those excluded groups separately, because bundling them into one number hides both your real competitive performance and your qualification problem. Always break the result down by deal size and segment, since a single blended figure usually averages away the finding that matters.
Do we need a competitive intelligence tool like Klue or Crayon?
Only if someone owns the function. These platforms start around $15,000 to $20,000 a year at the entry level and are genuinely good at what they do, but they make an existing practice bigger rather than create one. If no person has competitive intelligence in their job description, the alerts get muted and the cards go stale, and you have bought expensive shelfware. Start with CRM capture and an n8n monitoring workflow, then buy when you have an owner and more than five competitors that matter.
Who should own competitive intelligence at a 50-person company?
Usually product marketing owns the positioning and content, and RevOps owns the data model, the field validation, and the reporting. At 50 people neither is a full-time job, so the practical answer is a product marketer at 30% time with RevOps handling instrumentation. What does not work is making it everyone's responsibility, which is the same as nobody's.
How far back should we backfill competitor data?
Two quarters of closed-lost deals is the sweet spot. Beyond that, rep recall degrades badly and you are adding noise dressed as history. Run it as one timeboxed working session with bulk editing rather than a project, tag the backfilled records so you can separate them in analysis, and accept that your first honest quarter of data starts the day the stage gate goes live.
Where this leaves you
Competitive intelligence has a reputation as a product marketing discipline, and that is why it fails at companies where product marketing is one person with four other jobs. The part that makes it work is closer to plumbing: a properly modelled field, a validation rule at the right stage, a reconciliation against call data, and one dashboard.
Do that and the strategy conversations change character. Instead of arguing about which competitor is scariest, you look at a number, notice you lose 3 out of 4 deals above a certain size to one specific company, and decide whether to fix the gap or stop selling into that segment. Both are real decisions. Neither is available to a team running on anecdote.
If you want help wiring the capture layer into your CRM properly, that is the kind of thing we do in CRM and RevOps work. Get in touch and we will look at your current field structure and closed-lost history before anyone talks about tools.