How to Use Win/Loss Interviews for Competitive Intelligence

Almost everything in this cluster describes a competitor from the outside. Filings, adverts, review corpora and archived pages are all documents a rival produced, or documents produced about them, read by somebody who was never in the room. A win/loss interview is different in kind. It puts you in front of a person who watched your competitor sell, watched you sell, held both against a budget and a deadline, and then committed money to one of you.

That respondent is rare enough to be worth stating plainly. Analyst surveys ask people who have opinions about a market. Review sites ask customers about the product they bought. Neither has a sample of people who evaluated two named vendors against each other with real consequences attached. A win/loss programme is the only competitive research most companies can run where the comparison the whole business cares about has already been made by the respondent, at their own expense, before you called.

What it does not produce is a fact about your competitor. Everything you learn is a buyer’s account of what they were told and how it landed, filtered through the reason they are willing to talk to you. That is not a weakness to apologise for, because it is precisely the thing you cannot get anywhere else, but it does decide how the output should be written down. A win/loss interview establishes what a rival claims and which claims work. It never establishes whether the claims are true.

1. 1Decide which deals get an interview before you know how they ended. Write the selection rule first: every deal over a value threshold where a named competitor was present, won and lost alike. Choosing after the fact produces a programme that interviews painful losses and ignores easy wins, and that sample answers a different question from the one you asked.
2. 2Take the rep debrief first, and treat it as a hypothesis. Fifteen minutes with the seller before you contact the buyer gives you the deal timeline, the names, and the reason the rep believes they lost. Record that reason verbatim. It is not the answer; it is the claim the interview is going to test, and half the value of the programme is in the gap between the two accounts.
3. 3Have somebody other than the deal owner make the request. The ask comes from product marketing, research, or an outside interviewer, never from the person who worked the deal. Say plainly that it is a 25-minute conversation about how the evaluation went, that nothing said will affect any future commercial relationship, and that the seller will not be on the call.
4. 4Ask about the sequence before asking about the reasons. Walk the buyer through what happened in order: how the evaluation started, who joined it, what each vendor was asked to show, when the shortlist narrowed. Reasons offered before the sequence is established tend to be tidy after-the-fact stories. Reasons that emerge while somebody is reconstructing events are far closer to what happened.
5. 5Ask what the competitor said about you, and let the silence sit. This is the question the rest of the interview exists to earn. Ask it plainly, do not suggest an answer, and wait. The reply is a paraphrase of a rival's live positioning against you, delivered by somebody who has no reason to soften it, and it is the single most useful sentence the programme produces.
6. 6Record the exact words, then code them into a fixed scheme. Keep verbatim quotes, because the phrasing is what makes a finding usable later. Then code each interview against a fixed list of factors so that twenty conversations can be counted rather than merely read. A programme that produces only stories cannot show a trend, and a programme that produces only counts cannot explain one.
7. 7Close the loop in public, or the programme quietly dies. Report back what changed because of the interviews, naming the deal patterns rather than individual buyers. Sellers who see a battlecard claim rewritten from something a buyer told a researcher will introduce the next interview. Sellers who see nothing happen will stop, and the programme ends without anyone announcing it.

Ask a seller why a deal was lost and the answer arrives quickly and confidently: price, timing, a missing feature, a procurement process that favoured the incumbent. Ask the buyer the same question with a neutral interviewer on the call and a different account emerges, often about the evaluation itself rather than the product. Corporate Visions, describing what it calls an analysis of over 100,000 B2B deals, states that sales teams cite entirely different reasons for losing deals than buyers do 50 to 70 per cent of the time. It is a vendor’s own figure and the methodology is not published, so treat the range as an order of magnitude rather than a measurement. The direction is not in doubt to anybody who has run both interviews on the same deal.

The interesting question is why the gap is so wide, because the answer tells you what to do with it. Neither party is lying. They are answering from different vantage points, with different audiences, about different fractions of the same process.

The operational consequence is that the two accounts should be collected separately and kept separately. Recording the seller’s reason before the buyer interview and never overwriting it turns every deal into a small experiment: the reason predicted, the reason reported, and the distance between them. After twenty deals that distance has a shape. If your sellers consistently say price and your buyers consistently describe a demo that did not address their case, the finding is not about pricing at all, and no amount of discounting will move the number.

Most programmes are built out of losses, because a loss is what creates the urge to investigate. The result is a research function that only ever samples failure, and it misses the single most valuable conversation available: the buyer who heard your rival’s complete pitch, including everything they said about you, and then signed with you anyway.

That person has no reason to protect the competitor and no awkwardness about the outcome. They will tell you which claims the rival led with, which of your weaknesses the rival named, what evidence was offered, and which parts of the attack they found persuasive even though it did not change their mind. This is the closest thing to a competitor’s live battlecard that can be obtained honestly, and it arrives in the buyer’s own words with a date attached.

Three things only a win interview produces

• The attack that failed, and why it failed. A rival’s claim that convinces a lost buyer and a won buyer looks identical in a loss interview. Only the win interview shows you which counter-argument worked, which is the sentence that belongs in an objection handling document rather than a guess about what might work.
• What they nearly chose, and what nearly did it. Buyers who chose you will happily describe how close it was and what would have tipped them. That is a specification for the next release or the next price change, written by somebody who was prepared to pay.
• Which of your claims were checked, and how. A won buyer remembers the moment they stopped taking your word for something and went to verify it: a review site, a reference call, a trial. Knowing where verification happens tells you which surfaces are load-bearing in your category, and review sites are usually one of them.

There is a practical argument as well as an analytical one. Won buyers answer more often, reply faster, and are still commercially engaged, so a programme that includes them reaches a usable sample size months earlier than one restricted to losses. If the interview count is the constraint, and it usually is, this is the cheapest way to relax it.

A mature programme is not one interview repeated. It is a small set of distinct conversations, each answering something the others cannot, and each with its own reliability. The table above grades them; what follows is what each one is actually for.

1. The loss interview

The conversation everybody starts with. Its job is to reconstruct the evaluation in order: how it began, who joined, what each vendor was asked to demonstrate, when the field narrowed, and at what point the decision was effectively made even if the paperwork came later. Ask for the sequence before you ask for reasons. Reasons offered cold are frequently a tidy post-decision story; reasons that surface while somebody is reconstructing a timeline are far closer to what happened.

2. The win interview

The one described above, and the one to add first if your programme currently runs only losses. Same interviewer, same length, one additional question: what did the vendors you did not choose say about us, and what did you make of it.

3. The no-decision interview

Deals that ended with nobody buying anything are frequently the largest bucket in a B2B pipeline, and they are routinely excluded because there is no competitor to attribute them to. That is exactly the reason to run them: they describe the alternative you lose to most often, which is the status quo. Report them as their own category. Folding them into losses inflates your competitive loss rate and hides a problem that is usually about deadlines and internal sponsorship rather than about any rival.

4. The churn or switcher interview

Run 30 to 90 days after a customer leaves for a named competitor. The interview is looking for two things: the trigger event, which is rarely the reason given on the cancellation form, and what the competitor promised that made the cost of switching worth paying. Switching cost is the main defence most vendors have, so a rival who has found a way to cover it has found something worth knowing about. Several of these conversations read together are what turns a list of departures into a description of how a rival takes customers.

5. The evaluator versus the economic buyer

These are two different interviews and it is worth doing both on deals that matter. The evaluator ran the comparison and can tell you how the products differed in use. The signer negotiated and can tell you what was conceded at the end, which nobody below them saw. Programmes that only ever reach the evaluator systematically under-report commercial factors, then conclude that price does not matter much.

6. The recording and transcript of the interview

Ask for consent to record and take it seriously if it is refused. The reason to record is that the buyer’s exact phrasing is the durable asset. A summary that says the buyer found your onboarding heavy is a note; the sentence “we worked out it was six weeks before anyone in my team would be using it” is a competitive claim you can act on, and it does not survive paraphrase.

7. The evaluation scorecard or RFP the buyer used

Frequently offered if you ask at the end of a good interview. A scorecard is what an entire buying committee agreed mattered before any vendor demonstrated anything, with weights attached. Read across several of them and you are looking at what your segment buys on, independent of how any individual deal went.

8. The rep debrief, recorded separately

Fifteen minutes with the seller, before the buyer is contacted, capturing the timeline, the names, and their stated reason. Its reliability as an account of the buyer’s decision is low and it is graded that way deliberately. Its value is as the control: the prediction against which the buyer’s account is read.

9. What the buyer volunteers about the competitor’s commercials

Discount depth, contract length, ramped pricing, and whatever was added at the end to close it. Buyers offer the shape of this readily and the exact figure rarely, and the shape is usually enough: knowing a rival routinely concedes a second year at the original rate changes how you negotiate without needing their price list.

10. The reference calls the buyer took

Almost nobody asks about this and it is one of the highest-yield questions available. A competitor’s reference list is a curated selection of their happiest accounts, so the names tell you which segment they are most confident in, and what those references said went wrong tells you what a rival’s best customers still complain about.

11. A third-party interview programme

Buyers say harder things to strangers. An external firm also brings comparative data across its client base and removes the internal argument about whether the researcher had an agenda. The costs are real: a slower cycle, less product context in follow-up questions, and a per-interview price that limits volume. Most programmes that use both send the deals that will change a decision outside and keep the rest in-house.

What it costs

In-house, the real cost is calendar time rather than money: about 15 minutes for the seller debrief, 25 to 30 minutes with the buyer, and 30 to 45 minutes to write the interview up and code it. A programme running eight interviews a month is most of a day a week for one person, which is why programmes fail quietly when they are somebody’s fifth priority. Outsourced, pricing is normally per completed interview or on a retainer, and the number that matters when comparing providers is the completion rate rather than the headline price.

Who makes the request, and how

Not the deal owner. The request should come from a named person who did not work the deal, state a fixed length, say that the seller will not be present, and make clear that nothing said affects any future commercial relationship. Include your own name and employer up front. Beyond being the professional norm, it is what the recognised codes of practice for this work require, and an interview obtained without it is one you cannot describe later.

Consent, recording and what you promise

Ask before recording, and say what the recording is for and who will hear it. Where you promise anonymity, honour it in the reporting as well as the transcript, which in practice means reporting patterns across deals rather than quotes traceable to one identifiable buyer at one identifiable company. If you intend to share verbatim quotes internally, say so at the time rather than deciding afterwards.

Where the interviews come from

The selection rule should be written before you look at any outcomes: for example, every closed deal above a value threshold where a named competitor appeared, won and lost alike, plus every churn to a named competitor. Picking deals afterwards is how a programme ends up interviewing only the losses that stung, which is a sample assembled to confirm what somebody already thinks.

Everywhere else in this cluster the research is a document. Here it is a conversation with a named person, and that changes the risks completely. Nothing here is exotic, and all of it comes down to the same test: could you describe exactly how you obtained this, to the buyer, on the record, without anybody being uncomfortable.

• Introduce yourself and your employer before anything else. Introducing yourself as a researcher without naming your employer is the version of this that people talk themselves into, and it is the one that ends a programme when it surfaces. Disclosure is the baseline expectation in the established codes of ethics for competitive work, and it costs nothing: buyers agree to these conversations knowing full well who is asking.
• Never encourage a buyer to breach a confidence. Proposals, quotes and evaluation materials frequently carry confidentiality terms. Asking is fine. Accepting the first refusal is required. Suggesting a way around the obligation, or reassuring somebody that nobody will know, converts an ordinary interview into inducement, and the fact that the buyer volunteered it afterwards will not help you.
• Do not treat the interview as a sales call in disguise. If you promised no commercial consequence, that promise binds the follow-up too. Routing a lost buyer’s interview into a re-engagement sequence is the single fastest way to destroy the response rate, and word travels inside an industry faster than most people expect.
• Handle the recording as personal data, because it is. A transcript identifies a named individual, their employer and their opinions. Get consent to record, limit who can access it, say how long you will keep it, and delete it when that period ends. Where you promised anonymity, make sure the report cannot be reverse-engineered from deal size and date.
• Keep incentives small, uniform and disclosed. An honorarium is normal research practice. It stops being normal when it is large enough to buy a view, offered only to the buyers likely to say something useful, or framed as payment for particular information rather than for time.

• Anything about deals you were never in. The method samples evaluations that included you, so a rival who never had to face you at all leaves no mark on the sample whatsoever. Proxy: their published customer logos and case studies, and the segments their hiring is pointed at.
• Whether what the competitor claimed is true. You learn what was said and whether it worked, which is not the same as whether it was accurate. Proxy: their own documentation and changelog, which is written for customers who will find out.
• Any rate or share. Interviews explain outcomes; they cannot count them, and a set assembled from memorable deals will mislead badly if treated as a sample. Proxy: your own deal records, which is where every ratio in a win/loss report should come from.
• The competitor’s actual price. Buyers give ranges, structures and impressions, and their recollection of a number they saw once is unreliable even when they are trying to help. Proxy: transacted prices in your own closed-lost records, and public contract awards.
• What the buying committee said when no vendor was present. Your respondent reports the meetings they attended and summarises the rest. Proxy: the written scorecard or RFP, which is the one artefact produced by those meetings.

Run interviews continuously rather than in quarterly campaigns. The window that matters is the 30 days after a decision, so a programme that batches its interviews into the last two weeks of a quarter has already lost the competitive detail on two thirds of its sample. Book from a standing rule and a standing calendar slot: a fixed number of interviews a week, selected by the rule you wrote in advance, whoever is available.

Re-read the whole coded set twice a year rather than only reading each interview once. A claim that appeared in one conversation in March and three more by August is a rival changing their pitch, and that pattern is invisible if every interview is filed after it is written. Re-open the set early when a competitor raises money, changes their pricing page, or appears in deals where you had never met them before. Keeping the resulting picture of each rival in one place, updated as interviews land, is the job a competitor profile does; the interviews supply the only section of it that comes from buyers.

The buyer’s account does not expire, but the thing they were describing does. An interview from March is an accurate record of how a rival sold in March: the claims they led with, the price they held, the integration they promised was weeks away. By the time the same finding reaches a seller in a battlecard six months later, the rival may have shipped that integration, moved their entry tier, or dropped the argument entirely. Nothing in the transcript tells you which. The interview is a fixed point, and the competitor keeps moving away from it.

That is the expensive failure in this work: not a missing interview, but a well-evidenced finding quoted with confidence long after it stopped being true, by somebody who trusts it precisely because it came from a real buyer. Holding the underlying facts steady between interviews is the standing job that competitive intelligence software took over. Flares watches each rival’s pricing, product and messaging and dates every change, so a finding from March can be tested against this week before anybody repeats it to a customer. The conversation itself stays manual, and always will. Nothing observes what a buyer was told in a room they invited nobody into, and no quantity of published material replaces asking the one person who sat through both pitches.