.webp)
.png)
Win loss analysis is the practice of finding out why you actually won or lost a deal, rather than why the rep thinks you did. It matters because the field in your CRM marked "lost reason" is filled in by the person with the least incentive to be objective, usually in a hurry, usually from a dropdown.
It is normally sold as a research project: a third party interviews your lost prospects, once a quarter, and delivers a report. That model works and most teams never start it. There is a faster version, and you already have the raw material for it.
The bottom line: the real reason a deal was lost was said out loud while the deal was still live, which is why the recordings matter as much as the interviews. An interview recovers what the buyer remembers months later, the conversations hold what they actually said at the time, and the cheapest place to start is the source you already own. Analyse the wins alongside the losses too: losses only tell you what to stop doing, and the difference between the two lists is the finding you are looking for.
What win loss analysis actually tells you
Three things, in ascending order of value.
The decision criteria you were actually judged on. Not the ones in the RFP. The two or three that mattered when the buying committee argued about it in a room you were not in.
Where in the process you lost. A deal lost at the technical evaluation and a deal lost at the budget approval are different failures with different fixes, and both get recorded as "lost to competitor".
Who actually decided. Most lost deals reveal a stakeholder nobody on your side had mapped. That finding changes your qualification criteria rather than your pitch, which is why it is the most valuable of the three.
What it does not tell you is whether your product is good. Win loss analysis measures the selling, not the thing being sold, and teams that conflate the two end up rebuilding features in response to a messaging problem.
Two sources, and what each one is good for
Most people think of win loss analysis as interviews. That is one of two sources and it is the lossier one.
| Post-decision interviews | The deal's own conversations | |
|---|---|---|
| What you get | The buyer's considered explanation | What was actually said, when |
| Strength | They say things to a third party they would never say to you | No memory decay, no retrofitted narrative, full coverage |
| Weakness | Recalled weeks later, rationalised, and most people decline the interview | Only covers what reached a conversation with you |
| Coverage | The deals that agree to talk, typically a minority | Every deal, won and lost |
| Cost | A research project, internal or outsourced | Already paid for, if the conversations are recorded |
The interview is where you learn that your champion never had the budget authority they implied. The conversation record is where you discover they said so, in passing, on the second call, and nobody noticed.
Use both if you can. If you are going to use one, start with the one you already own. If you want the interview side run properly by a specialist, Gartner maintains a market overview of win loss analysis providers.
Why price is almost never the real reason
Price is the most frequently recorded lost reason in B2B and one of the least often true. Three reasons it dominates the dropdown.
It is the easiest thing for a buyer to say. "You were too expensive" ends a conversation politely. "Your rep never convinced our CFO this was a priority" does not.
It is the easiest thing for a rep to log. It attributes the loss to something outside their control, and nobody in the pipeline review will challenge it.
It is sometimes a value problem wearing a price costume. A buyer who understands the return does not describe the price as too high; they describe the budget as not yet allocated, which is a different and fixable thing.
The pattern underneath most of these is not price at all. It is indecision: the deal did not go to a competitor, it went nowhere. HBR's work on losing sales to customer indecision is the best treatment of this, and it reframes a large share of what teams record as competitive losses.
Which is why a win loss analysis that only reads the CRM field confirms what the CRM field already said. The finding has to come from somewhere the rep was not summarising.
How to run win loss analysis in six steps
One quarter's deals, two weeks of work, repeatable after that.
- Pick fifteen closed deals, not fifty. Ten lost, five won, all from the last quarter, all in your core segment. Fifteen is enough to see a repeated pattern and small enough that you will finish.
- Include the wins. Non-negotiable. Losses tell you what to stop; wins tell you what to repeat, and the difference between the two lists is the actual finding.
- Read the conversations, not the notes. For each deal, go through what was actually said: the objections raised, the competitors named, who attended which meeting, what was promised. The notes are a summary written by someone with a view.
- Tag against a fixed list. Decide your categories before you start: no decision, lost to competitor X, budget not allocated, wrong stakeholder, feature gap, timing. Inventing categories as you go produces a list nobody can count.
- Interview three or four buyers. Two losses, two wins, and have someone other than the deal owner do it. Thirty minutes each. You are checking whether your tags match what the buyer thinks happened.
- Write one page with three changes on it. Not a report. Three things that will be different next quarter, each with an owner.
The step teams skip is the second one. A loss-only analysis produces a list of things to fear, which changes behaviour in the direction of caution rather than in the direction of winning.
The questions worth asking in a win-loss interview
Six questions, and the order is deliberate: open wide, then narrow.
- Walk me through how the decision got made internally. Who was involved?
- What were you trying to fix? Has that changed since?
- When did you first feel this was likely to go one way or the other?
- What almost made you choose differently?
- What did our team get wrong, or not understand about your situation?
- If you were advising us, what one thing would you change?
Two rules make the difference. The deal owner does not run the interview, because buyers soften everything in front of the person who sold to them. And nobody defends anything: the moment the interview becomes a rebuttal, the useful part is over.
HubSpot has a longer list of win-loss questions if you want more to choose from. Six is usually the limit of what fits in thirty minutes with room for follow-ups.
Where the findings should land, and the mistakes that waste them
An analysis that ends in a slide deck has failed. Findings have to land in the four places where behaviour actually lives.
- Battlecards. A competitor pattern belongs on the battlecard within the week, not in next quarter's training.
- Qualification criteria. A recurring missing stakeholder becomes a question you cannot skip, and a stage you cannot exit without the answer.
- Objection responses. A recurring objection becomes a prepared answer, tested on calls. The ones that kill deals are usually never stated out loud, which is exactly why you need the conversation record to find them.
- The deal review agenda. If indecision is your top loss category, your deal reviews need a question about the cost of the buyer doing nothing.
Four mistakes waste the work. Reading only the CRM lost reason, which gives you back what you put in. Only analysing losses, covered above. Running it once, which gives a snapshot of one quarter's competitive landscape and no trend. And letting the deal owner tag their own deals, which is how "lost on price" becomes 60% of your categories.
If the analysis points at execution rather than at product or pricing, that is a sales performance diagnosis and it has its own method.
Where Praiz fits in
The expensive part of win loss analysis is not the thinking, it is the retrieval: going back through months of conversations to find what was said about budget, about competitors, about who decides. Done by hand, that is where the project dies.
Praiz removes that step. AI agents work on every call and meeting, video, phone and in person, and turn what was said into structured data: the competitors named, the objections raised, the decision criteria expressed, the stakeholders present, the risk signals. Three families cover it: agents that generate, agents that score each conversation against your own grid, and agents that track competitors and objections deal by deal.
For this exercise, that changes the unit of work. Instead of listening to fifteen deals, you filter a field: every deal where a given competitor was mentioned, every deal where budget was raised as an objection, every deal where the economic buyer never joined a call. The pattern that takes two weeks to find by hand is a query.
What teams measure: a 20% improvement in win rate once these signals are acted on, ten times more weak signals detected, fifteen times more structured customer feedback, and every call scored instead of a sample. The agents are configurable, because your decision criteria are not generic: start from the agents that analyse every conversation or build your own against the categories you decided on in step four.
And a buyer's question worth asking any vendor in this space: who configures it with you. Most hand over an empty screen. The Praiz team sets the agents up with you, which matters here more than usual, because the value of this analysis depends entirely on extracting the right categories rather than generic ones.
See it in action
Stop guessing why deals were lost
Praiz AI agents extract the competitors, objections and decision criteria from every conversation, so a win loss analysis becomes a filter instead of a fortnight.
Frequently asked questions
What is win loss analysis?
The practice of establishing why deals were actually won or lost, rather than relying on the reason a rep selected in the CRM. It draws on two sources: interviews with buyers after the decision, and the conversations that led up to it.
Done properly it tells you the criteria you were really judged on, where in the process you lost, and which stakeholder nobody had mapped.
How many deals do you need for a win loss analysis?
Around fifteen closed deals from the same quarter and the same segment, roughly ten losses and five wins. That is enough to see a repeated pattern.
Waiting for a statistically clean sample is the most common reason this never gets done. A pattern across ten deals that you act on beats a perfect study you never finish.
Why is price the most common lost reason?
Because it is the easiest thing for a buyer to say and the easiest thing for a rep to log. It ends the conversation politely and attributes the loss to something outside the rep's control.
The pattern underneath is often indecision rather than price: the deal went nowhere rather than to a competitor. That distinction changes the fix entirely.
Who should run the win-loss interviews?
Anyone except the person who owned the deal. Buyers soften their answers in front of the rep who sold to them, which removes most of the value.
A colleague from another team works, and a third party works better, because buyers say things to an outsider they would not say to the vendor at all.
.png)
.png)
