Skip to content
Win-loss analysis

Why did we lose the deal? A buyer interview guide

Published

A lost deal lands in the CRM under “missing feature.” Product sees another roadmap request. Sales wants a stronger demo. The founder wonders whether a discount would have helped. One label is already pulling the team toward three different fixes.

Before committing budget, ask the buyer to reconstruct the decision. What were they trying to do? What did they evaluate? When did your product stop being a serious option? The answers help you choose between product work, clearer proof, different targeting and changes to the buying experience.

Start with one decision, invite comparable wins and losses, reconstruct the timeline and separate the buyer’s account from your interpretation. Then choose a small change you can check. Buyer win-loss interviews give you evidence about specific evaluations. A few willing respondents cannot tell you how every prospect thinks.

This guide includes an outreach message, interview questions and a filled analysis example. The named cases come from public customer accounts checked on 22 October 2026. Most are published by the interview providers themselves; they illustrate practical problems without independently proving business results. The dates below identify publication, not the dates of the buying decisions.

Choose the decision before the interview questions

Write the decision you need to make in a sentence. “Understand our losses” leaves too much room for an interesting conversation that changes nothing. “Decide whether to improve the integration demo or investigate a missing workflow for our mid-market buyers” gives the interviews a useful boundary.

Choose a product, use case and sales motion. Keep an enterprise procurement cycle separate from a self-serve trial unless comparing those motions is the question. Renewal interviews can reveal why customers stay or leave, but their experience includes implementation and ongoing use. Mixing them with new purchases muddles what buyers knew when they chose.

Agree who will act on the findings. A founder can own qualification changes. A product lead can investigate feasibility. A sales leader can change demonstration practice. If nobody can spend time on the resulting work, narrow the question before recruiting.

Seismic: a cleaner CRM still left a missing perspective

In Klue’s Seismic case study, published 16 July 2024, Valerie Bonaldo’s team first improved incomplete CRM data. That made sellers’ explanations more visible. They still lacked the buyer’s view of the evaluation and its effect on the decision. The account describes a win-loss program begun in 2020 with the business now called Klue Win-Loss, formerly DoubleCheck.

For your preparation, keep two fields: what the seller reported and what the buyer later described. A cleaner seller account is useful preparation, but it cannot stand in for the buyer’s experience. Preserve both when they disagree.

The source also describes starting with a hypothesis the team wanted to prove. Avoid that aim in your own brief. Write an alternative explanation beside your preferred one and ask what would make you abandon it. A demo problem, an unavailable capability and a project with no funding lead to different decisions.

Select buyers without letting enthusiasm choose the sample

Build the invitation list from eligible opportunities before asking sellers for their favorites. Sales can flag a genuine contact restriction or correct an outcome. Ask for the reason behind exclusions so uncomfortable losses do not disappear unnoticed.

Use a recent period in which your product and commercial offer were reasonably consistent. Approach buyers while they still remember the evaluation, after the outcome is settled enough that your invitation will not feel like a negotiation. There is no universal interview deadline. Record the interval between decision and interview so a remembered evaluation from last quarter is distinguishable from last week’s.

Include three outcomes:

  • Won: what convinced buyers and which disadvantages they accepted.
  • Lost to another option: how your offer was evaluated and eliminated.
  • No purchase or postponed: what prevented a decision and what happened to the project.

Treat unknown outcomes separately. A closed-lost CRM record does not establish that a competitor received an order.

Within these groups, consider buyer role, company size, use case and evaluation depth. A technical evaluator knows different things from the budget holder. An early rejection contains different evidence from a completed trial. You may purposefully recruit a neglected group to understand it better, but the resulting mix will not represent your actual pipeline proportions.

Maintain a recruitment log with eligible, invited, accepted and completed interviews. Note refusals and nonresponse without guessing their motives. Review coverage as interviews arrive. If only friendly winners respond, change recruitment or qualify the conclusions; do not rename that group “the market.”

Illustration of equal invitation groups in three colours beside a reply group dominated by one colour

Illustration: different response patterns can leave the completed interviews unlike the buyers you invited. The envelope groups are instructional, not a measured research sample.

Balanced invitations do not remove nonresponse bias. People with time, strong feelings or a continuing relationship may be easier to recruit. You cannot fix missing perspectives by giving the available quotes equal space in a presentation.

Recruit for learning and protect the buyer’s information

Choose someone who did not sell the deal where possible. Product marketing, a colleague from another product or an outside interviewer can provide distance. Each still brings assumptions. Brief them on the product and the decision without handing over a verdict to extract.

Consensus: the sales leader struggled to get candid answers

Clozd’s Consensus account, published 3 April 2023, describes CRO Rex Galbraith trying to interview buyers himself. He found it difficult to ask the right questions, take useful notes and obtain candid feedback from people who had dealt with him directly.

The same account says later research helped Consensus refine unsuitable customer profiles and reconsider how it introduced a new use case. That expands the question beyond what feature might recover a loss: was the buyer ready for the proposed way of working?

The public case does not disclose those profile details or compare interviewers under controlled conditions. It supports a practical caution about relationship pressure, not a claim that paid outsiders are unbiased. If the founder must conduct the interviews, disclose the relationship, invite criticism and keep the seller off the call. Record that limitation when sharing findings.

A copyable invitation

Adapt the bracketed fields before sending:

Subject: A short conversation about your evaluation of [product]

Hi [name],

We’re reviewing how buyers evaluated [product] for [use case]. Would you be willing to spend 25 minutes walking through what happened and what mattered to your decision?

This is a research conversation. We won’t pitch or try to reopen the sale. [Name and relationship to our company] will conduct it.

We’ll explain who can see the notes before we begin. Recording is optional and we’ll ask separately. You can skip any question. Please avoid information your employer expects you to keep confidential.

If you’re interested, choose a time here: [booking link]. You can also reply with a time that suits you. It’s fine to decline.

Thanks, [name]

Use a duration you can honor. Send a restrained reminder if appropriate, then stop. Keep any thank-you payment independent of the answers and compatible with the buyer’s employer policies. Procurement and public-sector contacts may face restrictions. Avoid incentives tied to returning to the sales process.

ZoomInfo: remove the scheduling obstacle

A Product Marketing Alliance article, published 5 November 2019, quotes Andrew McCotter-Bicknell at ZoomInfo, described there as powered by DiscoverOrg. He used Calendly linked to product managers’ calendars so willing interviewees could choose a time without extended scheduling exchanges.

Apply that lesson to the gap between agreeing and attending. Offer actual interviewer availability, a clear time zone and a way to reschedule. Keep accepted and completed interviews as separate counts. The article provides no denominator for its scheduling-success claim, so it cannot establish a response-rate improvement. Easier booking also does nothing to establish that a product manager will ask neutral questions.

Before starting, explain the purpose, who will receive notes or recordings, where they will be stored and when they will be deleted. Ask permission to record separately from permission to participate. Offer notes only if recording is declined. Follow applicable recording rules and your company’s privacy requirements.

Do not promise anonymity if a deal’s details make the person recognizable internally. Explain that risk and offer a restricted summary. Give participants a contact for corrections or withdrawal requests and explain any limits once findings have been combined or shared. Internal research permission does not authorize a public testimonial, a sales clip or an identifiable quote in a company-wide channel.

Agree access and retention before uploading anything to a transcription or AI service. Confirm your organization permits that service for buyer information. Remove unnecessary personal details and confidential commercial material. If a buyer starts sharing a rival’s protected bid, redirect to the criteria they used rather than collecting the document.

Reconstruct the buying timeline before probing loss labels

Prepare a short fact sheet: product, use case, dates, participants, known outcome and available evaluation materials. Mark the seller’s explanation as seller-reported. List your hypotheses separately so you can see which ones the interview supports or contradicts.

The prompts below are suggested questions, not excerpts from actual interviews. Use them flexibly. You want enough chronology to locate a decision, with room to follow something unexpected.

StageNeutral openingUseful follow-up
TriggerWhat was happening when you started looking?What were you doing before that?
ShortlistHow did the options you considered change?What made an option worth evaluating?
EvaluationWhat did you need to establish before choosing?How did you test it and who took part?
Turning pointWhen did our option stop being a serious contender?What happened around that point?
ApprovalHow was the final decision approved?Which parts did you take part in directly?
OutcomeWhat happened to the project afterward?What did you choose and what disadvantages did you accept?

For winners, replace the elimination question with when your option became the preferred choice and what remained unresolved. Separate what they knew when choosing from what they learned after implementation. Later satisfaction or disappointment can color their account of the earlier decision. For postponed projects, ask what prevented the next step. Leave space for the buyer to say the project ended altogether.

Ask one question at a time. Follow the buyer’s words: “You mentioned implementation. What work did you expect your team to do?” If the answer is broad, ask for an event: “Can you walk me through the meeting where that became a concern?”

Separate direct experience from what someone heard later. “Were you in that discussion?” is often more useful than another broad question about why the company chose. A technical evaluator may know the tests well and know little about final budget approval.

Avoid presenting your preferred answer. “Did our missing integration cost us the deal?” tells the buyer what to confirm. “Which workflow were you trying to complete?” lets them describe the requirement. Save specific clarification until after the unprompted account.

Do not correct a mistaken product belief during the research conversation. Ask what they saw, heard or tried and where that understanding came from. If they ask a support question, offer a separate follow-up with permission. Reopening the pitch changes the conversation you invited them to have.

Finish by summarizing the sequence in their terms. Ask what you have misunderstood and whether an important participant or event is missing. A corrected summary is more valuable than a neat closing answer.

Separate purchase criteria from complaints and perceptions

Buyers can dislike something without rejecting a product because of it. They can also accept a serious drawback because another requirement matters more. For each issue, ask when it entered the evaluation, who cared about it and what happened next.

Classify the evidence at the level your decision requires:

FindingWhat to establish before acting
Capability was absentRequired workflow, product version and actual availability at evaluation time
Capability was unknownWhat the buyer encountered and which people received the explanation
Proof was unconvincingThe test or evidence needed and what the evaluation supplied
Implementation was unacceptableMigration work, skills, dependencies and deadline
Economics or terms failedTotal commitment, payment timing, approval constraints and alternatives
Purchase stoppedFunding, priority, ownership and whether anyone bought anything

A pricing complaint deserves its own investigation. Ask which costs and terms the buyer compared and how they worked through them. Subscription price, migration effort, minimum commitments and payment timing may affect different approvals. Sometimes the available budget really cannot cover your offer. Do not turn every affordability limit into a messaging problem.

When a buyer describes migration as the obstacle, our guide to SaaS switching offers helps separate the vendor’s offer from the work still left with the customer. Ask which of those tasks affected this evaluation.

If you need to examine the offer itself afterward, our SaaS pricing-page analysis guide helps identify what was publicly stated. Keep those page facts separate from the buyer’s private quote and approval process.

SmartBear: the feature existed but the buyer missed it

Clozd’s SmartBear case, published 5 May 2022, describes a pilot spanning two products. Jennifer Lowenthal’s team suspected that a different buyer type cared more about a particular capability than their traditional buyer did.

An interview then surfaced a buyer who believed SmartBear lacked that capability, even though the company already offered it. According to the account, SmartBear changed discovery and demos with the technical team supporting sales.

The actionable distinction is between the product’s capability and the buyer’s understanding during the evaluation. Before commissioning a build, check availability at the relevant date and what evidence reached the people involved. The public case names neither the feature nor the buyer and supplies no transcript. It does not establish exactly where the explanation failed or show that this explained every loss.

Illustration of a product brochure containing a capability omitted from the adjacent demonstration sequence

Illustration: a capability can exist in the product while remaining absent from the demonstration a buyer sees. This generic product is a teaching example, not a reconstruction of SmartBear’s demo.

SurveyMonkey: ask which integration and why it mattered

In Klue’s SurveyMonkey case, published 8 June 2025, Claire Brunvand describes earlier interviews that stopped at “integrations” as a loss reason. She says later research explored which integrations mattered, why they mattered and the business value behind them.

The next step is to reconstruct the required workflow. An integration name alone leaves product guessing about direction of data movement, required records, permissions and acceptable manual steps. Ask the buyer to describe the task from beginning to end, who performs it and what fails if it cannot be completed.

These are suggested probes; the source does not disclose the integration names or underlying workflows. After the interview, inspect documentation and a working test where possible. Our guide to reading competitor integration pages can help locate public clues. A partner logo cannot establish that a particular buyer tested or needed the connection.

Analyse accounts without manufacturing certainty

Write a short summary immediately after each interview while its sequence is fresh. Keep exact excerpts separate from paraphrases. Link quotations to the recording or notes permitted by the buyer. Mark uncertain recollections and unresolved dates.

Code each issue for its role: purchase requirement, deciding factor, accepted disadvantage or general complaint. Distinguish themes the buyer volunteered from topics you introduced. An issue they never mentioned remains unknown unless you explored it; silence does not show it was unimportant. Allow several issues per deal. A complaint appearing repeatedly within one interview still comes from one account. When several people from one buying committee participate, count people and deals separately. Their perspectives are useful related observations; they are not several independent losses.

Compare accounts within the chosen cohort. Look for counterexamples: a winner who shared the same concern, a loss where the proposed fix was already present or a buyer who never reached that evaluation stage. These cases help define when a finding applies.

When a seller says price and a buyer says implementation, preserve both. They may describe different stages, different stakeholders or two constraints that operated together. Check the proposal and timeline. With permission, speak to another participant if the disagreement would change your decision. Do not give the most senior speaker an automatic casting vote.

A handful of interviews can expose a specific problem worth checking. Report the number interviewed, eligible population, outcome mix and important gaps alongside any theme counts. Describe “three interviewed buyers raised this” rather than “buyers want this.” Avoid percentages that suggest precision beyond your recruitment method.

A filled decision record using the SmartBear case

This is an instructional synthesis of the published account, not SmartBear’s internal analysis. No transcript, feature name, buyer quotation or test result has been invented. The proposed checks and owner roles below are our recommendations.

FieldFilled example
DecisionInvestigate whether buyers need better evidence of an existing capability before requesting engineering work.
Published buyer accountClozd reports that a buyer rejected SmartBear believing a relevant feature was absent.
Published company accountThe capability existed; SmartBear changed discovery and demonstrations.
Evidence available to this guideA vendor-hosted customer case published 5 May 2022. No underlying recording or demo materials.
Working interpretationThe evaluation left at least one buyer with an incorrect understanding of capability.
Alternative explanationsThe capability may have been unclear, unavailable to that buyer on the relevant terms or insufficient for the specific workflow. These possibilities are unverified.
Missing evidenceFeature and plan at the evaluation date, buyer’s required task, demonstration sequence and participants.
Proposed next stepVerify availability, then add a task-specific demonstration and ask a future evaluator to explain what it established.
Proposed ownerProduct marketing with a product specialist; sales owns use in relevant evaluations.
Decision limitThis public account supports investigating a proof gap. It does not justify estimating its frequency or forecasting recovered revenue.

The proposed demonstration check measures comprehension. If buyers still describe a missing capability after seeing the task performed, investigate terminology, access and whether you demonstrated the right task. If they understand it and still reject the product, preserve their reasons instead of declaring the research finished.

Choose a research scope your team can carry

Internal interviews cost preparation, scheduling, conversation time, note checking and analysis. Reserve capacity for all of them. As an illustrative starting allowance, set aside 20 minutes for preparation, 25 for the conversation and 30 for checking notes and coding one interview. Add separate time for recruitment and cross-deal analysis. Log the actual time after the first calls and reduce the batch if review is falling behind. These allowances are planning suggestions, not measured industry benchmarks.

A founder-led pilot gives quick product context but introduces relationship pressure and competes with selling time. A colleague outside the deal adds distance while still needing training. An outside interviewer may provide capacity and practice, but needs product context and oversight. Ask to review the guide, coding approach and evidence access before buying a large program.

Written follow-ups suit buyers who decline calls and can clarify a specific point. They leave less room to explore an unexpected answer. Record the collection method rather than treating a brief email and a detailed interview as equivalent accounts.

Choose an initial batch your interviewer and decision owner can review, then schedule a coverage check before inviting more. Continue when a relevant group is missing, accounts conflict on a decision-changing fact or new interviews keep changing the proposed action. Narrow the scope when several products or buyer types produce incompatible stories.

Pause when you have enough evidence for a bounded, reversible test and record which questions remain open. That is a decision about the next action, not a claim that you have heard every possible explanation. Exhausting the invitation list does not establish research saturation. If you cannot recruit the relevant people, report that gap and postpone an expensive conclusion.

LaunchDarkly: somebody still has to act

In Clozd’s LaunchDarkly account, published 1 December 2022, Amelie Sutsakhan describes interview evidence helping internal discussions move away from finger-pointing. She also says acting on the findings remained LaunchDarkly’s responsibility.

Use that division of work when budgeting. Name the person who can change a demo, qualification rule or product investigation. Give them the supporting account and any accounts that disagree. The source does not supply a specific implemented fix or independently measured business effect, so it supports this ownership lesson without proving that research alone improved results.

Use the first month to test the process

Week one: define and prepare. Choose the decision, cohort and decision owner. Pull the eligible deal list, agree contact restrictions and write the consent terms. Prepare the interview guide and a recruitment log. Review the guide for questions that smuggle in your preferred answer.

Week two: recruit and listen. Send invitations across the outcome groups. Conduct initial interviews and check whether the prompts produce sequences rather than general opinions. Adjust unclear questions without changing the learning objective halfway through the cohort.

Week three: compare and investigate. Review coverage, pursue missing perspectives and check disputed product facts. Draft the decision record with counterexamples. Share only the buyer material your consent terms allow.

Week four: choose a bounded action. Correct a verified factual error immediately if its remedy is clear. Trial a clearer demonstration when the evidence suggests a proof gap. Before committing to expensive product work or changing pricing, seek corroborating deals and check feasibility and business economics. The evidence threshold should reflect the cost and reversibility of the decision. Assign an owner and a review date. Record what you expect to observe and what result would make you reconsider.

Treat this as a planning sequence, not a promise that enterprise buyers will answer within a month. Recruitment may take longer. A useful first-month result can be a clearer question and an exposed evidence gap.

Measure process quality first: coverage of relevant outcomes, usable timelines, unresolved contradictions and whether anyone acted. Then check the specific change. Did evaluators understand the demonstrated capability? Did qualification identify the readiness issue earlier?

Revenue and win rate move with deal mix, pricing, competition and sales execution. A before-and-after improvement does not isolate your interview program’s effect. Keep comparison periods and definitions consistent, disclose concurrent changes and avoid selling a small pilot as causal proof.

When Competitor Tracker & Co. fits

Competitor Tracker & Co. fits when your buyer interviews need public-page context alongside the private accounts you collect.

  • “I need to check what competitors were claiming while buyers evaluated us.”
  • “I want relevant pricing and product-page changes without checking every site myself.”
  • “I need to separate a public claim from what a buyer actually saw or tested.”
  • “I want enough context to choose a useful follow-up question.”

We do not recruit buyers, conduct your interviews or prove private purchase motives. If you need confidential deal research or someone to run the program, arrange that separately.

Our AI agents watch competitors’ public pricing, product and positioning pages, ignore routine page noise and send a Monday report of relevant changes. Keep those observations beside your interview findings, with their dates and limits intact.

Use the decision record to choose which public change needs a buyer follow-up and which can wait.

If buyer decisions need more context, see a sample dossier or start a case with Competitor Tracker & Co.

— C. T. Lucky