A win-loss interview earns its keep when the questions ask about behaviour and sequence. Ask what the buyer did, in what order, who else was in the room, and what they had to write down internally to get it signed off. Twenty questions will fill a thirty minute call. Four of them do most of the work, and they are marked in the table below.
That design comes out of a 1977 psychology paper. Richard Nisbett and Timothy Wilson, writing in Psychological Review, laid out four identical pairs of nylon stockings in front of 52 shoppers and asked which pair was the best quality. The right-most pair was chosen over the left-most by a factor of almost four to one. No shopper ever mentioned position. When the researchers asked directly whether position had affected the choice, virtually all of them denied it, "usually with a worried glance at the interviewer".
They weren't lying. They had no access to the thing that actually moved them, so they produced a reason that sounded right. Your buyers do the same. Ask a lost prospect why they went with the other vendor and you'll get an answer inside two seconds, every time, and it will be a story assembled after the event. Most of the question guides you will find on page one of Google are built almost entirely out of why questions. That is the flaw this bank is designed around.
What questions do you ask in a win-loss interview?
Twenty, ordered the way the deal actually happened, so the buyer is recalling a sequence instead of defending a verdict. The middle column is the part most guides leave out. A question is only worth its slot if you can say what it tests, and if you cannot say what it tests, cut it and buy yourself two more minutes.
The third column matters more than it looks. Vague first answers are the norm in this format, and the follow-up is where the call is either won or wasted.
| # | Question | What it is actually testing | Follow-up when the answer is vague |
|---|---|---|---|
| 1 | Walk me through the day you decided this had to be solved. What happened? | Whether a trigger event existed or the need drifted in. Trigger events tell your outbound team which signals to watch for. | "What would have happened if you had left it another six months?" |
| 2 | Who raised it first, and who did they take it to? | The real entry point into the account, which is rarely the job title at the top of your target list. | "What did they say to get it taken seriously?" |
| 3 | Before you spoke to any vendor, what did you do? | The pre-vendor research path, and where your content needed to be and was not. | "What did you read or watch, and who did you ask?" |
| 4 | How did our name first come up? | The actual acquisition route, as opposed to whatever your attribution model recorded. | "Who said it, and what did they say about us?" |
| 5 | What else was on the list, including building it yourselves? | Your true competitive set. It usually differs from the one on your battlecards. | "Which of those got a call, and which got ruled out from the website alone?" |
| 6 | Who joined the evaluation after that first conversation, and why? | How the buying group assembled and who holds a veto. | "What did each of them need to see before they would sign off?" |
| 7 | What did you have to write down or present internally to get this approved? | The business case in the buyer's own language. The most reusable artefact in the whole call. | "What was the headline of that document?" |
| 8 | What was the first thing that gave you doubts about us? | The earliest point of friction, caught before it hardened into the official reason. | "What happened immediately before that, and who was on the call?" |
| 9 | What did you compare us on, in the order you compared them? | Evaluation criteria and their ranking. The order is worth more than the list. | "Which of those would have killed it on its own?" |
| 10 | What did the demo change your mind about? | Whether the demo did any work at all, or simply confirmed a decision already made. | "What did you expect to see that you did not see?" |
| 11 | What number went into the business case, and where did it come from? | Whether your value story survived contact with their finance function. | "Who checked that number, and did they push back on it?" |
| 12 | What did the vendor you chose do in the first week that we did not? | A behavioural difference you can copy, as opposed to a perceived one you can only argue with. | "When exactly did that happen, and what did it look like?" |
| 13 | Where did our pricing land against what you had expected before you saw it? | Price against expectation, which is far more useful than price against a competitor. | "What had set that expectation?" |
| 14 | Who on your side was against this, and what was their objection? | The internal blocker your rep never met and never heard from. | "Did anything change their mind, or did it stay unchanged?" |
| 15 | What was the last thing that had to be true before you signed? | The real closing condition, which is often procedural and almost never appears in the CRM. | "Who had to say yes to that?" |
| 16 | If the budget had vanished halfway through and come back a year later, which of us would you have called first? | Preference with the outcome stripped away. This is the question that catches a justification built after the fact. | "Why that one?" |
| 17 | What did we claim that you were not sure we could actually do? | The credibility gap inside your own marketing. | "What would have made you sure?" |
| 18 | What do you know now that you did not know during the evaluation? | The gap between the promise and the delivery, from the only people who can see both ends of it. | "What surprised you most?" |
| 19 | If you were running our sales process, what would you change first? | Gives the buyer permission to be blunt about the process without criticising a person by name. | "At which stage exactly?" |
| 20 | What have I not asked that I should have? | Whatever your framework missed. It is also the only question that improves the question set itself. | Say nothing and wait. The pause does the work. |
Question 12 flips on a win: what did we do in the first week that the others did not. Question 18 works on both sides, and on a won deal it is the closest thing you will get to a product roadmap written by a customer.
The four that change anything
Sixteen of those twenty give you context. Four of them produce a decision somebody has to act on by Friday.
- Question 7, the internal document. The sentence a buyer wrote to get budget released is the sentence that belongs on your homepage. They pitched you internally in their own language, and this is the only question that hands you that language verbatim. It costs nothing to collect.
- Question 9, criteria in order. Every vendor knows their evaluation criteria. Almost none know the ranking. Ranking is what tells you which slide moves to the front of the deck and which qualification question moves to the first call.
- Question 16, the counterfactual. Nisbett and Wilson's shoppers could give you a reason and could not give you the cause. Asking who they would call back takes the outcome out of the question, so what comes back is a preference instead of a defence of a decision already made. When it contradicts the stated loss reason, believe this one.
- Question 20, the open door. Every research programme has a blind spot shaped exactly like its own questionnaire. This is the only question that finds yours.
If you have fifteen minutes rather than thirty, run 1, 7, 9, 16 and 20, and stop there.
Who should run win-loss interviews?
Anybody except the rep who owned the deal.
Roger Tourangeau and Ting Yan reviewed the survey methodology evidence on this for Psychological Bulletin in 2007 and concluded that misreporting on sensitive topics is largely situational. Respondents, in their words, "edit the information they report to avoid embarrassing themselves in the presence of an interviewer or to avoid repercussions from third parties". A buyer explaining to a salesperson why they rejected that salesperson is the textbook version of that situation. The buyer has nothing to gain from candour and an awkward ten minutes to lose.
The rep also walks in with an objective the researcher does not have, which is to come out of the call not having been the reason. That objective leaks into the wording of the questions, which is the subject of the next section.
Two owners work. Someone internal who has never carried the number on that account, which in practice means product marketing or competitive intelligence. Or an outside analyst, which costs money and gets the highest response rate, because buyers will say things to a stranger with no stake that they will not say to a supplier. We build ours around a single named analyst who owns every finding in the report, for the same reason a byline sits on a newspaper article.
What a bad win-loss interview sounds like
Neither exchange below is a transcript. Both are constructions, written to show the mechanism at work.
Exchange one, the leading question.
Interviewer: So at the end of the day it came down to price, didn't it?
Buyer: Yeah, I mean, price was definitely a factor.
Interviewer: Understood. Anything else?
Buyer: No, I think that was mostly it.
That deal is now logged as a price loss and somebody will convene a pricing meeting about it. What actually happened is a verb effect. Elizabeth Loftus and John Palmer measured the shape of it in the Journal of Verbal Learning and Verbal Behavior in 1974. Students watching the same car crash film estimated the speed at 40.8 mph on average when the question used the word "smashed", and 34.0 mph when it used "hit". In a second experiment with 150 students, asked a week after viewing, 16 of the 50 who had received the "smashed" question reported seeing broken glass in the film, against 7 of the 50 who had received "hit". There was no broken glass in the film. The question had edited the memory.
The repaired version asks question 13 instead. "Where did our pricing land against what you had expected before you saw it?" Now price has to compete for the answer rather than being handed it.
Exchange two, the rep who lost the deal.
Rep: Is there anything I could have done differently?
Buyer: Honestly, no. You were great throughout. It was just timing on our side.
Rep: That's good to hear. Keep us in mind for next year?
Buyer: Absolutely, definitely.
Four minutes of nothing, and a renewal date in the CRM that nobody will honour. The repaired version removes the rep and opens with question 8. "What was the first thing that gave you doubts about us?" It presupposes doubt, which is safe, because there always was some, and it gives the buyer somewhere to put it that does not require insulting anyone who is listening.
How many interviews before the findings hold?
Nobody has run a saturation study on B2B win-loss interviews specifically. Say that out loud before you quote any number at a client. What does exist is a solid body of work on the point at which qualitative interviews stop producing new themes, and the read-across is close enough to be worth having.
| Study | What it analysed | Where saturation landed | Read-across for win-loss |
|---|---|---|---|
| Guest, Bunce and Johnson, Field Methods, 2006 | 60 in-depth interviews, coded in ten cumulative rounds of six | 73% of all codes appeared in the first six interviews, and 92% by twelve. Of the 36 codes that ended up being applied most often, 34 were already visible in the first six. | The themes that end up mattering most show up early. Six interviews is enough to form a hypothesis, and nowhere near enough to reprice a product on. |
| Hennink, Kaiser and Marconi, Qualitative Health Research, 2017 | 25 in-depth interviews, tested against two different definitions of saturation | Code saturation at nine interviews. Meaning saturation at 16 to 24. | Nine calls tells you what the issues are. Understanding why they bite takes closer to twenty. Plenty of programmes stop at the first number and report it as the second. |
| Hennink and Kaiser, Social Science and Medicine, 2022 | Systematic review of 23 empirical tests of saturation | Saturation within 9 to 17 interviews where the population was fairly homogeneous and the objective narrowly defined. | Segment before you count. Nine interviews spread across four industries and two deal sizes is four samples of two, and it will hold nothing. |
None of those three studied buyers. They studied health behaviour, and I wouldn't put them in front of a client as anything more than the closest available evidence. Read with that caveat, they still say something firm. The answer sits in the nine to twenty range per segment, and a programme running four interviews a quarter across an entire customer base is producing anecdotes with a chart on top.
The ranges we plan against at ORRJO come off that table. We haven't run our own saturation study and I'm not going to imply otherwise. Nine to twelve completed interviews inside one segment before a theme gets written up as a pattern. Closer to twenty before anyone moves a price on it. Those are the ranges we plan against, not promises, and they widen as the buyer population gets more mixed.
Should you interview the deals you won?
Yes, and the split matters more than most programmes allow for.
Losses tell you what to fix. Wins are the only control group you have. Without them, every finding from a loss is uncontrolled. You hear "your implementation timeline worried us" from four lost deals and conclude the timeline is the problem, when the same four won deals would have told you the timeline worried them too and something else outweighed it. A loss-only programme overstates the importance of friction as a matter of arithmetic, because friction is the only thing it collects.
Wins carry the opposite bias. A customer who has just signed has every reason to be generous, which is why question 19 exists: it gives them a socially safe way to be critical. Run the full twenty on both sides and resist the pull towards a shorter script for the happy calls.
Interview the deals that never reached a decision as well. Losing to a competitor and losing to nothing are different failures with different fixes, and only one of them is a competitive problem. Our win-loss analysis service treats no-decision as its own cohort for exactly that reason.
How soon after the decision should you interview?
Fast, and the argument for speed is about memory rather than diary logistics.
Klue's 2025 Win-Loss Trends Report, built on 313 leaders involved in win-loss, reports that 40% of deals go through win-loss analysis on average, and that 70% of the deals analysed had closed within the month before. Klue sells win-loss software, so read the framing accordingly. The number underneath it is still a fair description of the practice: most of what gets analysed is analysed while it is fresh.
The Loftus and Palmer broken glass result is why freshness matters. One word inside a question changed what people reported having seen a week later. Every internal debrief your buyer has sat through since the decision, and every justification they have given their own boss, is post-event information doing that same job on the memory you are about to ask for. Interview at week two and you get the decision. Interview at month four and you get an account of the decision that has been rehearsed four times.
What I could not verify
Three things I went looking for and will not be quoting.
The most repeated claim in this category is that sales reps are wrong about the loss reason a large share of the time, usually put at 80% or 85%. Every version I traced ran back to a vendor blog carrying no sample size and no stated method. The claim is plausible and it matches what we find in client CRMs. No primary source publishes it, so it is absent from this article, and you should treat it as an argument wherever you meet it.
Gartner's work on the size of B2B buying groups gets quoted constantly in guides like this one. It sits behind a bot wall that would not open for me, and I am not going to relay a figure through somebody else's blog post. Question 6 gets you your own number for your own market, which beats an industry average anyway.
There is no published benchmark for how many lost buyers will agree to a win-loss interview. Not one. Anyone quoting you a response rate for this is quoting their own book, which may well be a reasonable guide and is not a benchmark.
Where this fits
The question bank is the cheap part. The expensive part is running the calls often enough, and consistently enough, that quarter three is comparable with quarter two, which is why most in-house programmes produce two good reports and then quietly stop. If you are weighing up doing it internally against buying it, we keep an independent comparison of B2B win-loss providers that includes the ones we lose to, and a shorter buyer research interview programme for teams who want the method without waiting on closed deals.
ORRJO Intelligence starts at £2,500 a month and runs win-loss as a standing programme with a named analyst on every report.
One last thing, and it's the failure I see most often. A team runs twenty good interviews and then compresses them into a slide that reads "price and timeline". The value was sitting in the verbatim sentence the buyer used in their own business case, and summarising is precisely the operation that deletes it. Keep the sentences.