How should we structure win-loss interview design to uncover the specific objections that lose deals?

Two ways to build the program: neutral third party versus internal research

Every win-loss design collapses into one structural fork before any question gets written: who conducts the interview. The two options are a neutral third-party research firm and a dedicated internal researcher, and they produce measurably different data from identical deals.

The third-party option buys three things you cannot easily manufacture in-house. First, interviewer neutrality — a buyer will tell a stranger with no commission exposure that the discovery call was shallow, and will tell the AE who lost the deal "it came down to price." That single substitution is the largest known distortion in the discipline; internal-AE-conducted interviews inflate the pricing code by roughly two to three times relative to neutral interviews of the same deals. Second, recruiting reach — a vendor that books buyer interviews for a living has scripts, incentive mechanics, and persistence patterns tuned to the hardest population in the program, which is lost-deal buyers. Third, methodological maturity — a locked guide, calibrated coders, and an inter-coder agreement practice that an internal team has to build from zero.

The internal option buys different advantages. An internal researcher understands the product deeply enough to hear a vague complaint and probe to the specific missing capability behind it, rather than recording "wanted better reporting" and moving on. They can chase a technical thread a generalist interviewer would let pass. Their marginal cost per interview is dramatically lower once the role exists, which means the program can afford volume — and volume is what produces saturation inside individual segment cells. They also live inside the routing forums, so the distance between a finding and a roadmap conversation is a hallway rather than a quarterly deliverable.

The failure modes are equally distinct. A third-party program fails by distance: the readout arrives as a polished PDF, the vendor does not sit in the roadmap review, nobody in the room owns the findings, and the transcripts become an expensive archive. An internal program fails by capture: the researcher reports to a function with a stake in the answer, the taxonomy quietly drifts toward categories that flatter the current roadmap, and the loss sample skews toward the friendly buyers who were easiest to book.

There is a third structure that most mature programs converge on, and it is worth naming explicitly rather than treating as a compromise. The hybrid assigns the loss and no-decision interviews — where neutrality carries the most weight and recruiting is hardest — to the third party, while an internal researcher runs win interviews, owns the taxonomy and the code book, performs synthesis, and drives the routing map. The org buys neutrality exactly where bias is most expensive and keeps ownership exactly where organizational distance is most expensive. The hybrid also solves the succession problem: vendor relationships end, but the taxonomy, the code book, and three years of coded transcripts stay in-house.

One decision that is *not* a fork: whichever structure you pick, the interviewer must never be the rep who carried the deal. That is not a preference, it is the load-bearing control that the entire instrument rests on. A program that lets AEs interview their own lost buyers has not built a bias-correction instrument at all — it has built a mechanism for laundering rep self-attribution through a buyer's polite agreement.

How to decide between them

The decision is driven by four inputs, and they are worth evaluating in a fixed order because they gate one another.

Start with routing capacity, not budget. If product, pricing, and enablement have no bandwidth to act on findings for the next two quarters, neither option is correct — the program will produce a readout that decays into cynicism, and the second-year budget conversation will be brutal. Confirm there is a forum, an owner, and a decision SLA for each taxonomy category before you evaluate vendors. A program with excellent interviews and no routing is strictly worse than no program, because it consumes credibility as well as cash.

Then check deal volume against cell saturation. Qualitative research reaches usefulness at thematic saturation, and field practice converges on roughly 12–20 interviews per persona-by-segment cell before themes stabilize. If your closed-deal volume cannot fill even two cells at that depth in a year, you do not have a vendor-selection problem — you have a scope problem, and the answer is a lightweight founder-led or PM-led variant rather than either formal option.

Then weigh the neutrality premium against the specificity premium. The question to ask is which error would cost you more this year: mis-attributing losses to price when the real driver is discovery quality, or capturing a product-gap finding too vaguely for a product manager to write a ticket from it. In a competitive, feature-mature category where the loss narrative is contested internally, neutrality wins. In a technically deep product where the gaps are subtle and the loss reasons are already broadly agreed, internal specificity wins.

Then price it. A pilot at 20–40 interviews runs roughly $15K–$45K fully loaded and supports one segment directionally. A standard program at 50–90 interviews runs roughly $35K–$110K and supports two to three segments with quarterly readouts. A comprehensive program at 100–180 interviews runs roughly $110K–$280K and can hold a full segment matrix with monthly competitive signal. Those bands include vendor fees, internal analyst time, incentives, and tooling — quoting only the vendor invoice understates true cost by a meaningful margin.

A note on the timing of this decision: make it before you write the guide, not after. The guide, the consent language, the recruiting message, and the incentive mechanics all differ slightly depending on who is calling, and retrofitting a guide written for an internal voice onto a third-party interviewer produces an awkward instrument that neither party runs well.

The specific objections you are trying to surface, and what each one costs

The reason interview design matters at all is that the objections that actually lose deals are systematically invisible to the people closest to them. Rep-logged loss reasons disagree with neutral buyer interviews of the same deals a large majority of the time, and the disagreement is directional rather than random — price is the only loss reason that does not implicate the rep's own discovery, demo, or follow-through, so it absorbs the blame that belongs elsewhere. This is not dishonesty; it is ordinary motivated reasoning under social and financial pressure, and it is exactly what a structured instrument exists to correct.

Six objection families carry nearly all the signal, and the taxonomy should lock at that number. Fewer than five and the categories are too broad to route to a specific owner. More than eight and inter-coder agreement collapses, because analysts genuinely disagree about which bucket a quote belongs in.

Product gap. The buyer needed a capability you lack or have at insufficient depth. The coding bar is specificity: "missing SSO" is unusable, while "could not provision via SCIM, which their IT mandated for every SaaS purchase" is roadmap-ready. The standard is that a product manager should be able to write a ticket from the code without a follow-up call. Sub-tags beneath this category — integration, depth, missing module — add routing precision without touching the locked six.

Pricing and packaging. The category most corrupted by self-attribution, so the bar is highest here. A pricing code requires the *buyer* to have named price as the deciding factor, plus detail on whether the problem was absolute cost, model fit (per-seat versus consumption), tier packaging, or procurement-stage discount friction. A vague "too expensive" gets coded unverified and excluded from the trend. This discipline matters because pricing changes are expensive and hard to reverse — you do not want to restructure tiers on the strength of folklore.

Sales experience. Discovery depth, responsiveness, demo relevance, multi-threading, trust. This is where the price myth most often gets corrected, and it is also the fastest category to act on: a product gap needs a quarter of engineering, while a discovery-quality miss can be trained into a team in a thirty-day enablement cycle.

Competitive parity. A specifically named rival out-featured, out-positioned, or out-referenced you. Code the competitor name *and* the exact advantage — a parity code that says only "the other vendor was strong" is useless to a battlecard author. Reference customers matter disproportionately here: a peer-company reference call that de-risks the competitor in the final stage is a decisive event the rep frequently never learns about.

Implementation and risk. Fear of painful onboarding, hard data migration, weak support reputation, or organizational change-management cost. This surfaces in the buyer's internal risk conversation rather than on any sales call, which is why reps rarely see it. It gets mis-coded as a product gap constantly; the distinction is that implementation risk is about the *path to value*, not the value itself, and it routes to customer success rather than engineering.

Internal politics. The champion left, the sponsor got reorganized, budget froze, or the business case never cleared the status-quo bar. Left uncoded, this category makes an org believe it has a product problem when it actually has a compelling-event problem.

Beyond the taxonomy, several silent killers deserve naming because they are structurally outside rep visibility: procurement objections raised in the final ten percent of the cycle after the rep's last call; security and compliance rejections decided in a meeting the rep was not invited to; sponsor handoff failures where the champion's departure killed the deal and the rep logged it as "timing"; and internal build-versus-buy debates the rep never knew were running. None of these appear in a CRM loss code. All of them appear reliably in a well-run fifty-minute interview.

The cost of not capturing them is not neutral — it is negative. An org without a structured program does not have missing loss data, it has *corrupted* loss data steering the roadmap, the discount policy, the battlecards, and the ICP. Budget approval for a program is therefore not an additive spend against a clean baseline; it is the correction of an existing and expensive error.

Concrete numbers behind each option and each design choice

Respondent mix. Modern B2B purchases are committee decisions, and the design should target three voices per studied deal: the economic buyer who explains why the money moved, the champion who explains how the process actually unfolded, and the technical or security influencer who explains the gates and disqualifiers. You will not always get all three; a realistic achieved rate is 1.6–2.0 respondents per studied deal. Design for three and treat a single-voice deal as incomplete, because a champion-only program under-weights procurement and security objections while an economic-buyer-only program misses the process detail that drives playbook fixes.

Outcome mix. Allocate roughly 30–35% of slots to closed-won, 35–40% to competitive losses, and 25–30% to no-decisions. That third category is the one most programs skip, and for many categories a large share of forecasted pipeline dies there. The no-decision failure mode — weak business case, no compelling event, status-quo bias — is completely different from a competitive loss and needs its own interview design. Wins are not padding either: a "shallow discovery" loss code only means something if you can demonstrate that won deals had deeper discovery. Without the win baseline, every loss finding is an uncontrolled anecdote.

Response rates and the loss-sample trap. Winners agree to interviews at roughly 35–55%; losers agree at roughly 8–18%. Left unmanaged, that differential over-samples wins by three to four times and hands leadership a picture materially rosier than reality. Four counters, in order of leverage: set loss-weighted recruiting quotas at the *final* mix you want and over-recruit losses three to four times to hit it, treating the quota as a contract term with your vendor; offer a modest honorarium or charitable donation, which lifts loss participation and is a rounding error against deal value; send a brief, gracious note from the CRO framed explicitly as "we want to learn, not re-sell," separate from the interviewer; and request the interview within two to three weeks of close, before the relationship goes cold.

Timing window. Interview within 60–90 days of close. Earlier than about three weeks and the loser is still disengaging and hard to book — and you catch both parties at emotional extremes, with fresh winners uncritically enthusiastic and fresh losers still annoyed. Later than 90 days and timeline recall collapses: buyers compress, reorder, and rationalize a sequence they can no longer actually remember, and a reconstructed timeline cannot support root-cause analysis.

Guide length. Twelve to sixteen core questions with branching probes. Longer forces a rushed, surface-level pass; shorter misses the timeline depth that makes the data credible. Train interviewers to treat the core questions as scaffolding and the probes — the "walk me through that" follow-ups — as the actual work.

Coding mechanics. One primary code per deal plus up to two secondary codes; forcing a single primary prevents the "everything mattered" cop-out and produces a clean rank ordering. Double-code 15–20% of transcripts with a second analyst, and treat inter-coder agreement below 80% as a signal that the definitions are too loose to keep coding. Lock the taxonomy for at least a year — changing categories mid-stream destroys trending, which is the entire payoff. Tag every code verified or unverified based on whether the buyer, not the rep, named it, and let only verified codes feed the trend.

Program economics. Rigorous programs are benchmarked at meaningful win-rate improvement over a twelve-to-eighteen-month horizon, and the mechanism is leverage rather than volume: one correctly coded and routed product-gap finding changes a roadmap decision affecting every future deal in a segment, not one deal. One corrected sales-experience pattern trains an entire AE team. That systemic quality is why the return per dollar is unusually high for a revenue-intelligence investment — and why a sub-scale program running under a dozen interviews per cell is genuinely dangerous rather than merely weak. It produces noise that looks like signal, and leadership acts on it with unearned confidence. If you cannot fund 12–20 per cell, narrow to fewer cells rather than thinning every cell.

Segmentation math. A three-by-three-by-three matrix of segment, persona, and outcome is 27 cells; saturating all of them would require several hundred interviews a year, well past most budgets. The discipline is deliberate cell selection — pick the four to eight cells with the most strategic uncertainty and the most pipeline at stake, saturate those, and mark the rest explicitly out of scope for the year. A program that covers every cell thinly saturates none.

Implementation details and sequencing

The guide is a narrative reconstruction, not a question list. The buyer's decision unfolded as a story in time, and the reliable way to recover accurate detail is to walk them back through it chronologically. Opening with "why did you choose the other vendor?" invites a rationalized answer — the same motivated reasoning that corrupts rep attribution, now running on the buyer. Five phases, always in this order:

*Timeline reconstruction.* What was happening in the business that started the evaluation, who first raised it, who got pulled in as it got serious, what deadline or event drove the clock. Low-threat factual recall that builds rapport.

*Vendors evaluated.* Who made the shortlist, who got cut early and what cut them, how the buyer first heard about each, and whether an internal build-it-ourselves option was on the table. This phase routinely reveals a competitive set that differs from what the CRM recorded.

*Decision criteria.* The top five must-haves in rank order, whether that order shifted during the evaluation, whether a single criterion became a dealbreaker, and which "nice to haves" turned out not to matter. This is the heart of the objection signal.

*Vendor selection.* Take them to the moment the decision was actually made and who was in the room. What did the winner do that others did not. Where specifically did you fall short. Would one point higher on any single dimension have changed the outcome.

*Post-decision reflection.* If they were advising you, what one thing should change. What nearly won or nearly lost it. Would they evaluate you again, and what would have to be true. Six months on, are they happy with the choice.

The front-loading is deliberate. A buyer who has spent ten minutes calmly reconstructing facts is far more candid when the hard question arrives than one ambushed with it in minute two. That rapport ramp is a data-quality control, not a courtesy.

Question craft. Open rather than leading — "how did pricing factor into your decision?" not "was our price too high?" Specific rather than abstract — "walk me through your second call with our rep" not "how was the sales process?" Behavioral rather than attitudinal — "what did you do in the 48 hours after the demo?" not "did you like the demo?" Single-barreled always. Silence-tolerant: wait five to seven seconds, because the best detail arrives after the pause. And non-defensive: hearing "your product could not do X" and answering "tell me more about that gap as you experienced it" instead of "actually, we can." Defensiveness ends candor instantly and permanently for that call.

Three light variants on one backbone. The win guide spends extra time on what made you the credible, low-risk choice and on the near-miss moment — wins almost always contain a defect hiding inside a success. The competitive-loss guide goes deeper on the criteria where the rival scored higher and on when and why they entered. The no-decision guide replaces vendor selection entirely with a business-case autopsy: what would have had to be true for *any* purchase, and what is the cost of the status quo they chose. Keep them as light edits of one backbone rather than three instruments, so cross-outcome comparability survives.

Opening and close are load-bearing. The opening states independence, that the call is to learn rather than re-sell, that there are no wrong answers, and that feedback will not be attributed by name — plus explicit recording consent naming who will hear it. The close asks "is there anything I should have asked about this decision that I did not?", which routinely surfaces a decisive factor no structured question reached, because a warmed-up buyer volunteers it. Both belong in the guide as fixed elements.

Interviewer calibration. A great salesperson is often a poor win-loss interviewer, because the instinct to handle objections and steer toward a close is precisely wrong. Run the first three to five interviews of any new interviewer with a senior reviewer listening to the recording against a rubric: did they lead, did they tolerate silence, did they chase the probe, did they defend the product. Cross-listen periodically when multiple interviewers are running, so style stays consistent enough for the data to be comparable.

Capture discipline. Record with consent and work from full transcripts, never interviewer notes — notes are themselves a bias filter, capturing only what the interviewer thought mattered in the moment. Transcripts make double-coding possible, let a second analyst find what the first missed, and supply the verbatim quotes the readout standard requires. Store them in a searchable repository; three years of transcripts only pay off if a battlecard author can pull every quote mentioning a rival in minutes.

Cadence. Run interviews continuously as deals close rather than in quarterly batches — rolling capture preserves the 60–90 day window for every deal, while a batch inevitably catches some at 30 days and others at 150. There is a second reason: continuous capture keeps the program within a quarter of current reality. A competitor ships in February, a rolling program interviews affected buyers in March and April and can flash the battlecard team in May. Synthesize and read out quarterly for the full program, with a monthly competitive flash for fast-moving battlecard updates, because the roadmap cannot absorb input faster than quarterly but competitive intelligence goes stale in weeks.

Detecting saturation. After each batch of five interviews in a cell, count genuinely new themes. Early on, every batch adds several; as the cell saturates, the curve flattens. When two consecutive batches of five add nothing new, redeploy budget to an unsaturated cell. One caution: an unsaturated cell can *look* saturated if the guide is too narrow or the interviewer is leading, so a flat curve should trigger one check on the instrument before the budget moves.

Routing, built before the first interview. Product gap routes to the head of product via quarterly roadmap review, roadmap call within a quarter. Pricing and packaging to the pricing lead or RevOps via the pricing committee, tier review within a quarter. Sales experience to enablement via the monthly sync, playbook update within 30 days. Competitive parity to product marketing via the monthly competitive sync, battlecard refresh within two weeks. Implementation and risk to customer success, process fix within a quarter. Internal politics to the CRO via QBR, deal-strategy update within a quarter.

Four artifacts every quarter. A roadmap reprioritization memo ranking gaps by frequency *and* deal value, so a rare gap on six-figure deals outranks a common one on small deals. A battlecard refresh per named rival with the specific advantage and the proven counter. An ICP refinement showing which segments win and lose and why — the highest-leverage output, because it changes what the org targets in the first place. Playbook updates tied to coded sales-experience misses, pushed to enablement inside 30 days.

Phased rollout. Weeks 1–3 are design: lock the taxonomy, build the sequenced guide, choose third-party versus internal versus hybrid, define the routing map and SLAs. Weeks 4–12 are the pilot: 20–40 interviews in one segment, validating the guide, calibrating coding, running the first double-coding check. Weeks 13–26 are scale: expand to the selected cells, hit 12–20 per cell, deliver the first quarterly readout and all four artifacts. Week 26 onward is the closed loop: route findings, ship the artifacts every quarter, track win-rate lift by segment, run the monthly competitive flash.

Operating failure modes to design against. The transcript graveyard, where interviews happen and nothing gets coded or routed — cause is no internal owner, fix is to name one. The stale taxonomy, tweaked every quarter so nothing trends — cause is well-meaning iteration, fix is a hard year-long lock. The optimistic sample, where wins dominate and leadership relaxes — cause is ignored loss-weighting, fix is enforcing the quota contractually. The readout nobody attends — cause is no pre-built forums, fix is building the routing map first. The vanity program that exists to be cited rather than to change decisions — cause is no win-rate attribution, fix is tying the owner's review to routed outcomes.

When the heavy program is the wrong build

Four conditions flip the answer, and a rigorous practitioner should recognize them rather than default to the full apparatus.

Very high deal volume at very low ACV. A self-serve or product-led business closing thousands of small deals gets better signal from product telemetry and a lightweight in-app exit survey. The cost per insight of a fifty-minute interview is prohibitive when each deal is tiny, and the purchase was rarely a committee evaluation worth reconstructing in the first place.

Pre-product-market-fit. When the founder is in every deal and total closed volume cannot fill a single cell, direct founder debriefs beat a formal program. The founder *is* the instrument at that stage. Graduate when there is enough closed-deal volume to saturate cells and enough routing capacity to act — typically well after the first institutional round.

A single dominant, already-agreed loss reason. If you genuinely lose most deals to one missing capability everyone already acknowledges, spend the money building it rather than interviewing buyers to re-confirm it. Win-loss earns its cost on *ambiguous* loss patterns; it is wasted on a consensus problem.

No routing capacity. Covered above, but worth restating as a stop condition: fix the forums first.

Even in these cases the principle holds — a revenue org must hear from the buyers who chose someone else. The lightweight variant is five to ten founder-led or PM-led loss calls per quarter, coded loosely to the same six categories so the data is forward-compatible when the heavy program eventually starts. The argument is against bolting enterprise-grade infrastructure onto a business that cannot yet use it, not against the discipline.

There is a mirror-image risk inside mature programs: over-engineering. Symptoms are a fifteen-category taxonomy nobody codes reliably, a twenty-five-question guide producing rushed interviews, and monthly full readouts the roadmap cannot absorb. Every element of the design should be sized to the cadence at which the receiving function can actually act. A program tuned for elegance rather than routed outcomes is just a quieter version of the vanity failure mode.

What win-loss cannot replace, and what it feeds

Win-loss sits next to three adjacent practices that are frequently confused with it, and knowing the boundaries prevents both duplicated spend and false confidence.

NPS and CSAT measure existing-customer sentiment after onboarding. They never reach the buyers who chose someone else, and they do not capture the competitive decision at all. They answer a retention question, not a selection question.

Conversation intelligence captures what was said on sales calls, which makes it superb for coaching and for verifying whether a discovery framework was actually run. What it structurally cannot capture is the procurement meeting, the security review, the internal build-versus-buy debate, or the executive hallway conversation — precisely the rooms where a large share of losses are decided. The two instruments are complements: call recordings tell you what your side did, interviews tell you what the buyer's side concluded.

Deal-desk and pipeline analytics measure internal process metrics — cycle time, stage slippage, discount depth. They will tell you a deal slipped; they cannot tell you the buyer picked a rival because a peer reference call de-risked them in week nine.

Downstream, the interview data has one further use that most programs under-exploit: the multi-quarter trend. Because the taxonomy is locked, a second-year program can show whether the shallow-discovery code is actually falling after the playbook change, whether a rival's parity advantage is widening, and whether a segment's win rate is responding to the ICP refinement. That converts win-loss from a quarterly anecdote generator into a management instrument, where leadership can watch a routed fix move the coded data over time. It is the whole reason the taxonomy must never be tweaked mid-stream — a year of locked discipline is what buys the trend, and the trend is what buys the program permanence.

One final note on ownership. Whichever structure you choose, there must be a single internal owner — usually in product marketing, RevOps, or competitive intelligence — who owns the taxonomy, the routing map, the quarterly readout, and the win-rate attribution. The owner does not need to conduct interviews. The owner needs to be the person whose performance review includes whether the findings changed a roadmap, a battlecard, and a playbook this quarter. Programs without that single accountable name reliably decay into archives.

Related questions

How many interviews do we need before the findings are trustworthy?

Roughly 12–20 per persona-by-segment cell, judged by thematic saturation rather than statistical significance. Watch the new-theme curve: when two consecutive batches of five add nothing new, the cell is saturated. Sixty interviews spread thin across eight cells has saturated none of them.

Should we interview won deals, or only losses?

Both. Wins reveal repeatable strengths worth protecting and, more importantly, supply the contrast class that makes loss findings interpretable. A "shallow discovery" loss code means nothing without evidence that won deals had deeper discovery. Target roughly a third of slots to wins.

Why can't our AEs just interview their own lost buyers?

Because buyers will not tell the person who lost the deal that the discovery was shallow — they say "price" to be kind. That single substitution inflates the pricing code by two to three times and hides the discovery, security, and politics objections the program exists to find.

What do we do about no-decision deals?

Interview them as a distinct third outcome with a modified guide. There was no vendor-selection moment, so replace that phase with a business-case autopsy: what would have had to be true for any purchase, and what is the real cost of the status quo they chose instead.

How fast should findings reach the field?

Two clocks. Competitive-parity findings go to battlecards within about two weeks via a monthly competitive flash. Roadmap, pricing, and ICP findings move on the quarterly readout, because the receiving functions cannot absorb input faster than that anyway.

FAQ

How long should each interview run?

Forty-five to sixty minutes. Under 45 you cannot walk the full five-phase timeline and still leave room for branching probes, so you get a survey rather than a reconstruction. Over 60 and senior buyers start declining the ask outright — the request itself becomes the barrier. A precise "45 minutes" is also more credible to an executive than a vague "quick chat," because it signals a professional program that respects their calendar.

What if the buyer refuses to be recorded?

Proceed unrecorded, but flag the transcript-less interview in the data set and do not use it for double-coding or verbatim quotes. Take structured notes against the five phases immediately after the call while recall is fresh. If refusals cluster in a particular segment or region, that is usually a consent-framing problem in the recruiting message rather than genuine reluctance — restate who will hear the recording and that nothing is attributed by name.

Can we use AI to code the transcripts?

Automated coding is reasonable as a first pass to suggest a category and pull candidate quotes, but the verified-versus-unverified distinction and the primary-code selection should stay human, because both depend on judgment about whether the buyer actually named the factor or merely mentioned it. Keep the same 15–20% double-coding check on machine-suggested codes and measure agreement against your human baseline before trusting the output in a trend.

How do we get loss interviews when nobody will take the call?

Four levers, roughly in order of effect: over-recruit losses three to four times to hit your target mix, offer a modest honorarium or charitable donation, send a gracious CRO note framed explicitly as learning rather than re-selling, and ask within two to three weeks of close before the relationship cools. Offering phone, video, and flexible scheduling also widens the funnel more than most teams expect.

Should the taxonomy ever change?

Not within a year. The entire value of a locked spine is that quarter three is comparable to quarter one, which is what lets you prove a routed fix actually moved the data. If a category genuinely does not fit your market, change it at an annual review and accept a documented break in the trend. Refine sub-tags beneath the six as often as needed — that adds routing precision without touching comparability.

Who should own the program internally?

A single named person, typically in product marketing, RevOps, or competitive intelligence, whose performance review explicitly includes whether findings changed a roadmap item, a battlecard, and a playbook this quarter. Ownership of the taxonomy, the routing map, the readout, and the win-rate attribution all sit with that role, regardless of whether a third party conducts the interviews.

Sources

• https://www.gartner.com/en/sales/insights/b2b-buying-journey
• https://hbr.org/2017/03/the-new-sales-imperative
• https://www.forrester.com/blogs/category/win-loss-analysis/
• https://www.gong.io/resources/
• https://klue.com/blog/win-loss-analysis
• https://www.clozd.com/resources
• https://openviewpartners.com/blog/product-led-growth/
• https://review.firstround.com/the-sales-playbook-that-helped-us-scale/
• https://www.pragmaticinstitute.com/resources/articles/product/win-loss-analysis/
• https://sloanreview.mit.edu/article/the-new-science-of-sales-force-productivity/

Related on PULSE

• [How should we design an objection taxonomy that avoids becoming a junk drawer?](/knowledge/q477)
• [When does a third-party win-loss vendor beat an internal research function?](/knowledge/q475)
• [What cadence of win-loss signal should trigger a roadmap or GTM pivot?](/knowledge/q476)
• [How do we turn competitive-parity losses into a take-out campaign?](/knowledge/q479)
• [How should win-loss findings reshape our ICP definition?](/knowledge/q480)
• [What belongs on a battlecard that answers real buyer objections?](/knowledge/q478)