Business

How to conduct win-loss analysis with buyer interviews

Stefan Weiss, Co-Founder & CEO

Stefan Weiss

How to conduct win-loss analysis with buyer interviews

To conduct win-loss analysis, define the buying decisions you want to understand, select a clear set of completed opportunities and interview buyers about how they reached their decision. Compare those accounts with the available deal records, preserve disagreements and turn the findings into specific changes you can investigate or test.

A win-loss study should explain more than which outcome occurred. It should help you understand the requirements, trade-offs, events and perceptions that shaped the decision. A reason selected in a CRM is useful context, but it is not a substitute for the buyer's account.

This guide is for founders, product marketers and commercial teams running a focused study. The workflow and worked example are editorial suggestions, not results from interviews with Weeve customers or a tested promise of higher win rates.

What you need before starting

Choose a decision the research should inform: improving a demonstration, clarifying product fit, investigating a recurring objection or deciding which product gap needs further work. “Find out why we lose” is too broad to guide a useful first study.

Have a bounded opportunity list, outcome definitions, basic deal context and a way to store authorised notes and source references. Identify an owner who can act on the findings. Decide who can contact and interview participants through your organisation's normal process, and make recording and access arrangements clear.

A spreadsheet can hold the working analysis. You need consistent questions and traceable evidence before you need a specialised system. Keep confidential deal information within the permissions and handling rules that apply to it.

Expected result: a study brief describing the question, included opportunities, evidence sources, limitations and intended decision owner.

1. Define the opportunity cohort and outcomes

Specify the period, product, customer group and stage of the buying process included. Record why those boundaries fit the question. Mixing different products, buyer groups or sales motions can make a single explanation difficult to interpret.

Agree what “won”, “lost” and “no decision” mean in your records. A buyer choosing another supplier differs from a buyer deferring the work or continuing with an existing process. Preserve those differences if they affect the question, even when your CRM uses a broader closed-lost category.

Keep the opportunity as the unit being compared. Several interviews about one opportunity can add perspectives, but they do not become several independent deal outcomes. Likewise, repeated comments from one interview do not establish that an issue affected several buyers.

Distinguish a research cohort from a business metric. A deliberately selected interview sample is not automatically the denominator for an organisation-wide win rate. Use the organisation's agreed reporting definitions for that separate calculation.

2. Include perspectives that could challenge the easy explanation

Seek both won and lost decisions relevant to the study question. Include no-decision cases when they are part of the problem. Keep a record of who was invited, who participated and which groups remain missing.

An interview with an enthusiastic new customer may be easier to arrange than a conversation with someone who chose another supplier. That difference can shape the material you collect. Describe the resulting limitation instead of presenting the participating buyers as a representative market sample.

Pragmatic Institute's guidance advocates hearing from won and lost opportunities and considering interviewer independence. Its specific sample prescriptions are practitioner recommendations, not a universal guarantee of statistical validity. A fixed split alone cannot make a self-selected interview sample representative. Pragmatic Institute: win-loss analysis practices.

Choose an interviewer who can listen without returning to the sale or defending every criticism. If the seller conducts the conversation, be explicit about that relationship and preserve it as context. Independence is a useful consideration, not proof that any particular account is unbiased.

3. Reconstruct the decision before asking for a verdict

Prepare a shared topic guide so that interviews remain comparable. Use it to explore the sequence, not to force every buyer through an identical script.

Start with the situation that triggered the search. Ask what the buyer needed to change, which options they considered and who influenced the decision. Then explore the evaluation and the point at which the choice became clear.

Useful prompts include:

  • What started the search for a different approach?

  • Which options, including keeping the existing process, did you consider?

  • What did you need to see before an option could remain under consideration?

  • Can you walk me through the moment that changed your assessment?

  • Which questions remained unresolved when you made the decision?

  • What made the chosen option workable for your organisation?

  • Looking back, what have I misunderstood about the decision?

When a buyer says “price”, ask what that meant in the evaluation. The issue might concern the amount, uncertainty about usage, the buying process or an unclear connection to the outcome. These are possible follow-up questions, not reasons to replace the buyer's answer with your preferred explanation.

Klue's interview guidance emphasises detailed accounts, natural follow-ups and resisting the urge to defend the product. Use those as interviewing principles; its commercial product claims and suggested questioning frameworks do not establish the quality of your own study. Klue: conducting win-loss interviews.


Separate orange and cream cloth speech bubbles face different directions with a clear gap between them.

4. Keep buyer accounts, seller accounts and records distinct

For each opportunity, maintain a short evidence record. Include the outcome and date, the participant's role, the points they raised, source references and unresolved questions. Mark exact quotations as quotations and paraphrases as paraphrases.

Keep the seller's interpretation alongside the buyer's account rather than replacing one with the other. A CRM reason may have been entered before the buyer explained the decision. An interview participant may also have seen only part of their organisation's internal discussion.

Where accounts disagree, write the disagreement clearly. “Seller recorded price; interviewed buyer described an unresolved implementation dependency” is more useful than immediately recoding the deal as a product failure. You may need another source or a clarification before drawing that conclusion.

Be careful with facts about competitors. A buyer's recollection of a competitor's feature, price or promise is evidence of their perception. Verify current product facts through that supplier's primary material before presenting them as established comparisons.

5. Compare reasons with their context intact

Create a small set of working categories that help answer the study question. Possible starting areas include task fit, implementation, evaluation experience, commercial conditions and internal decision constraints. Define what belongs in each category and keep a place for unclear or new material.

Apply the categories to specific statements, not whole buyers. One opportunity can involve several relevant considerations. Keep the original passage available so that a short category label does not erase the sequence or qualification that made it meaningful.

Count distinct opportunities separately from interviews and mentions. If you report a count, state the reviewed set and the unit. “Mentioned in three of the reviewed opportunities” has a different meaning from “three mentions”. Neither establishes how common the issue is among every prospect.

Look for cases that challenge a developing interpretation. If a capability was absent in both a win and a loss, it may still matter, but its absence alone does not explain the difference. Ask what other conditions changed the decision.

For a broader classification workflow, see how to categorise customer feedback. For this study, keep the completed opportunity and buying sequence visible rather than merging every comment into a general feature-request total.

6. Work through an example before presenting a pattern

Consider three fictional opportunities for an internal request-management tool. These are invented examples to illustrate analysis, not findings from tested products or interviewed customers.

Opportunity

Recorded outcome

Available account

What remains to check

A

Lost to another supplier

The buyer said a required approval handover remained unclear during evaluation

Which handover mattered, what was shown and what the selected supplier actually supported

B

Won

The buyer also asked about approval handovers, then accepted a documented manual route

Whether the manual work was acceptable because this buyer had different staffing or volume

C

No decision

The buyer described an internal ownership change before a final evaluation

Whether the project was deferred, abandoned or moved to a different team

A broad “approval problem” category could group A and B, but their outcomes differ. A useful tentative interpretation is that the operating conditions and acceptability of the workaround need closer investigation. “Approval features caused the losses” overstates the evidence.

C should not silently become another competitive loss. It may inform how the team identifies internal ownership, while contributing little evidence about the relative quality of competing products.

Keep the wording of the conclusion proportional: an observation from the reviewed opportunities, a possible explanation and the next check. A recurring theme can justify investigation without proving a causal effect.


Weeve recording selector showing synthetic example recordings with Product planning selected.

Genuine Weeve interface with demonstration recordings. Select the relevant study interviews for your own review; the entries shown are synthetic examples.

7. Use Weeve to review the selected interview sources

With permission to record and review the interviews, Weeve can retain the audio and transcripts on your Mac. Chat with recordings can help compare selected saved conversations and return source references for the passages used.

Select the relevant study recordings before asking for a cross-interview answer. A useful request is: “Across these selected buyer interviews, compare the stated decision criteria and turning points. Cite the passages, preserve disagreements and counterexamples, and identify which opportunities or details remain unclear.”

Inspect the sources and the reported coverage. Check the opportunity identities and important wording yourself before transferring an observation into the study record. An answer based on selected passages is not automatically a complete review of every interview or every buyer decision.

Weeve helps review what the recorded material contains. It does not automatically integrate the study with your CRM, verify competitor facts or establish why a market behaves a certain way. Internal buyer conversations that were never supplied remain outside the evidence.

Local AI workflows are available with downloaded models. If you choose an optional external provider for chat, relevant text is processed by that provider. Check the selected mode and permissions before using confidential interview material.

8. Turn findings into a specific next decision

For each proposed change, state the observation behind it, the interpretation and what the team wants to do next. Assign an owner and identify how the organisation will check whether the change addresses the problem.

For the fictional handover example, the next action might be to clarify the operating scenario in demonstrations and investigate the buyers for whom the manual route is unsuitable. That is a narrower, more testable response than promising to build a new feature because one lost buyer mentioned it.

Keep the distinction from sales call review: coaching a seller's behaviour and understanding a completed buying decision can inform one another, but they need different evidence. A good call does not guarantee a win, and a win does not show that every part of the evaluation worked well.

Finish with a clear account of the study's boundaries, the findings it supports and the questions still open. That gives product, marketing and sales a basis for the next decision without turning a small set of conversations into an unsupported market forecast.