Win loss analysis should tell you why deals close and why they die. Your CRM says you lose on price. Your CRM is wrong, and you already suspect it. Skip the CRM reason dropdown. Run fifteen-minute interviews with ten won and ten lost buyers each quarter, count the recurring themes, and attach one concrete change to each. The value is the pattern, not the form. Patterns beat anecdotes every time.
Why CRM reason codes lie
The loss reason in your CRM was chosen by the rep, from a dropdown, in four seconds, weeks after the conversation. Reps pick price because price is safe: it blames the market, not the call. Buyers offer price for the same reason. It is the objection everyone can agree to pretend about.
So the dropdown data accumulates into a chart that says price, timing, price, and nobody changes anything because nothing specific was learned. That is the theater: the ritual of analysis without the analysis.
Getting the interviews is easier than it sounds. Ask at the natural end: right after the yes, or two weeks after the no, when the sting has faded. Offer to share the anonymized findings. Buyers say yes to a short, specific request from a founder far more often than your calendar fears.
The fifteen-minute interview
Call the buyer. Three questions: what almost stopped you, or for losses, what finally stopped you; what were you comparing us against; what would have made the decision easier. Then follow up on the answers. Fifteen minutes is enough because you are after one specific thing, and specificity has a sound: a feature name, a meeting, a moment.
You will hear the real reasons in the first five interviews. Security arrived too late. The champion could not sell it internally. The competitor's pilot included an engineer and yours did not. None of those fit in a dropdown.
Who runs the interviews
Not the rep who owned the deal. Buyers protect the feelings of the person they rejected, and reps hear what lets them sleep. A founder works well. A marketer works. An outside interviewer works best at volume, but do not let the tooling outgrow the point.
The only requirements: the interviewer was not in the deal, and they are allowed to write down uncomfortable things. If your win-loss program cannot produce a sentence that stings, it will not produce a change either.
Count patterns, not stories
One interview is an anecdote. Ten is a pattern. Log each conversation as a few tags: late security, weak champion, missing integration, whatever the buyer actually said. Count them quarterly. The top three loss themes and the top three win themes are your go-to-market roadmap.
The count protects you from the loudest story. A vivid loss to a competitor will dominate a pipeline review for weeks. If it happened once in twenty deals, it is a story. If it happened eight times, it is a strategy problem with a name.
Close the loop or stop doing it
Every pattern gets one concrete change: security documentation sent before the first demo, a champion kit in every evaluation, the integration on the roadmap with a date. One change per theme, owned, dated. Then next quarter's interviews grade the fix.
That grading step, done in front of the whole team, is what separates a program from a project. If the changes are not happening, stop running the interviews and spend the hour on something honest. Win-loss analysis that does not change behavior is the one kind of theater your buyers will never applaud.
Frequently asked questions
What is win-loss analysis?
A structured review of why deals close and why they die, built from short interviews with the buyers themselves. Done honestly, it is the cheapest market research a startup has. Done as a CRM dropdown, it is fiction.
Who should run win-loss interviews?
Anyone except the rep who owned the deal. Buyers tell the truth to a neutral party; they protect the feelings of the person they rejected. A founder, a marketer, or an outside interviewer all work.
What questions should a win-loss interview ask?
Three: what almost stopped you, or what finally stopped you; what were you comparing us against; what would have made the decision easier. Fifteen minutes. The follow-up questions matter more than the script.
How many interviews do I need per quarter?
Ten wins and ten losses is enough for patterns to show. Five of each still beats none. If you cannot name the top three loss reasons after twenty conversations, the questions are wrong, not the sample.
What do I do with win-loss findings?
Count the themes and attach one concrete change to each: a pricing page fix, an earlier security review, a different discovery question. A pattern with no change attached is how programs turn into theater.