How to run win-loss analysis without a research budget

Win-loss analysis is the highest quality competitive intelligence there is, because it comes from the only people whose opinion converts into money. It is also mostly skipped, because it is uncomfortable.

You do not need a research firm. You need ten conversations and the discipline not to argue during them.

The setup

Who asks. Not the rep who worked the deal. Buyers are polite, and they will soften the answer to avoid a confrontation with someone they had a relationship with. Marketing, product, or a founder gets closer to the truth.

Who to ask. Ten recent closed deals, mixed: five lost, three won, two that went quiet without a decision. The last category is the most neglected and often the most revealing.

How to ask. A short email. Fifteen minutes, we are trying to get better, we will not sell you anything. Then do not sell them anything, ever, including three months later. Word gets around and it poisons the well.

Expect roughly one in four to agree. That is fine.

The five questions

Ask these, in this order, and then be quiet.

1. Walk me through how you decided. Who else did you look at?

Open, no leading. Their list of alternatives is the first finding, and it is frequently not the list you assumed.

2. What made you pick them, in the end?

Let them answer fully before saying anything. The first reason is the polite one. The second and third are the real ones.

3. Was there a point where we were ahead, or behind?

This locates the moment it turned. Often a specific meeting, a specific missing feature, or a price conversation that went badly.

4. What would we have had to do differently?

People are generous with this. They have usually thought about it.

5. Anything you expected from us that you did not get?

The catch all. This question produces the answers nobody wanted to volunteer, more often than the other four combined.

The rules during the call

Do not defend. The instant you explain why they were wrong, the interview is over. They will be polite for the remaining ten minutes and tell you nothing.

Do not correct factual errors. If they say your product does not do something it does, write that down as a finding. Their belief is the fact that mattered. Correcting them turns research into a sales call.

Ask "what do you mean by that" twice per call. Expensive, unclear, complicated and risky all mean different things to different buyers. The specific meaning is where the insight is.

Take notes, do not record. Recording changes what people say. Write up within an hour while it is fresh.

What to do with ten conversations

Put the notes side by side and count. You are looking for a repeated sentence, not a good quote.

Three kinds of finding come out.

Price. Sometimes real, usually a proxy. "Too expensive" almost always means "I did not see enough value to justify it", which is a story problem rather than a pricing problem. Test it: ask whether they would have bought at 20% less. Most say no, which tells you it was never price.

A missing capability. Real, and worth counting. If it appears in six of ten losses it belongs on the roadmap. If it appears once it is one buyer's requirement.

Process. Slow to respond, unclear proposal, too many people in the meeting, no answer to a security question. These are the cheapest to fix and the most commonly ignored because they are embarrassing.

The losses you never hear about

Here is the limitation you have to name.

Win-loss covers people who talked to you. The bigger number is people who never did, because you were not on the shortlist at all.

That group cannot be interviewed. You find them differently.

Search. Who ranks for your commercial terms. If three competitors sit above you, a lot of buyers never reached your name.

AI answers. Ask twenty buyer questions in ChatGPT, Claude, Gemini and Perplexity and record who gets named. Being absent here is a silent loss with no referrer and no record.

Directories and listicles. Whichever pages your category's buyers read. Absence from one that ranks first costs more than most ranking problems.

This is the invisible half of your loss rate, and it does not show up in any CRM. Treating win-loss as the complete picture is how a company concludes it has a product problem when it has a distribution problem.

The rhythm

Ten interviews twice a year is enough for most companies. Between rounds, one question in the CRM on every closed lost deal: who did they choose. That single field, populated consistently, is worth more than most competitive intelligence software.

Put the findings in the competitor profile you keep, dated. Six months later you will want to know whether the thing you fixed actually changed the pattern, and you cannot tell without the earlier version.