Win-loss program

Win-Loss Program: The Truth Source for Everything Downstream

Average B2B win rates sit around 19-21% (Ebsta x Pavilion GTM Benchmarks, 2025 edition, still the latest full dataset). Read as an information problem: roughly four of five worked opportunities end in a decision your company usually never investigates. The loss-reason field in the CRM does not count as investigation. It is filled by the person with the strongest incentive to externalize the cause, seconds after the sting, from a dropdown that forces one cause where there were three. Loss fields cluster on "price" and "timing"; buyer interviews of the same deals routinely surface trust, absent capabilities, a stronger champion on the other side, or a decision that was rigged before the first call. Every downstream system inherits whichever version you record.

A win-loss program replaces attribution folklore with three evidence lanes, run on a standing cadence: mine every decided deal, interview a sample of buyers one on one, and periodically put a group of customers in one room with the whole go-to-market team listening.

Lane 1: Transcript and Thread Mining (every decided deal)

The cheap lane, and with call recording in place, the one with no excuse. Within a week of any decision, mine the deal's full record: call transcripts, email threads, proposal versions, evaluation notes.

• Extract in the buyer's words: what they said about the problem, alternatives, money, timing, and internal politics. Verbatim quotes with timestamps, never paraphrases; a paraphrase is where wishful thinking re-enters.
• Reconstruct the decision timeline: when did the deal actually die or lock? It is almost never the dated loss event in the CRM; find the call where the energy changed and what preceded it.
• Contrast rep story vs record: write the rep's stated loss reason next to what the record shows. The recurring gap between the two, by rep and by cause, is itself a coaching output.
• Limits, stated honestly: mining only hears what was said TO you. The real reason often lives in the meeting you were not in. That is what Lane 2 is for.

Lane 2: Buyer Interviews (the sampled, deeper lane)

• Sampling: every large or strategic decision, a rotating sample of the rest, and both outcomes. Wins get interviewed too; a program that only studies losses learns half the causal model, and win interviews are where your actual differentiators (versus the ones on your slides) surface.
• Timing: 2-6 weeks after the decision (practice-based window). Sooner is raw; later is revisionist.
• Who asks: never the account's own rep. A neutral interviewer (founder on strategic deals, RevOps or PMM, or a third party) changes what buyers are willing to say; buyers soften the truth to the person they rejected.
• Protocol: 30 minutes, no selling, no defending, no correcting the record. One spine question: "walk me through the decision as it actually happened, from when you first felt the problem." Then follow the story: who was in the room, what almost changed the outcome, what the winning option did that mattered, what you did that annoyed or reassured them. Close with: "what would have had to be true for this to go the other way?"
• The discipline: when the buyer says something that stings, the interviewer writes it down and says "that is useful, say more." The program's value is measured by how often the findings are uncomfortable.

Lane 3: The Customer Arena (the group lane)

The customer arena is a lean service-management format (known in Dutch practice as the klantarena): a moderated session where a group of customers talks about their experience while the company's cross-functional team (product, marketing, sales, customer support) sits in the listening seats. Where Lane 2 reconstructs single decisions, the arena surfaces the pattern layer live, and it answers the question analytics never can. Usage data and funnel metrics tell you WHAT is happening; the arena is where customers tell you WHY, in each other's presence, building on each other's answers in a way one-on-one interviews cannot.

• Composition: 5-8 customers per session, mixed deliberately: happy accounts, accounts that almost churned, recent wins, and where you can get them, a lost evaluator or churned customer (one skeptic in the room raises everyone's honesty). Segment-pure per session; a group that shares a context talks specifics.
• The listening rule: the company side asks clarifying questions only. No defending, no roadmap promises, no selling. The moment someone from your side explains why the customer's experience was actually fine, the arena is over; it just keeps talking for another hour without telling you anything.
• The spine: what are you using it for, why that, where does it grate, what almost made you leave, what would you tell a peer who asked about us. Walk the journey chronologically (first contact, buying process, onboarding, daily use, support) so friction lands on a stage, not in a vague pile.
• The harvest is double-sided by design: every friction point gets tagged product or process. "The integration kept breaking" is a product finding; "we never understood the pricing until the third call" is a sales-process finding; "nobody contacted us for four months after go-live" is a post-sale-process finding. Arenas run this way are one of the few instruments that improve the product and the sales motion from the same hour of evidence.
• Cadence: one or two per year per segment is enough to matter; the constraint is customer attention, not your calendar. Run one after any strategy-level surprise in the Lane 1/Lane 2 findings; the arena is the fastest way to test whether an interview theme generalizes.
• The side effect is not a side effect: customers who spend an afternoon being genuinely listened to, alongside peers, leave more invested than they arrived. Do not abuse this by turning the arena into a marketing event; it works because it is not one.

Synthesis: From Stories to a Decision-Driver Taxonomy

Individual post-mortems are anecdotes; the program's output is the pattern layer.

1. Code every decided deal against a stable driver taxonomy (product capability, trust and proof, price-to-value, champion strength, competitive move, timing and budget reality, process failure). Multiple drivers per deal, weighted primary/secondary; single-cause coding rebuilds the dropdown problem you are escaping. Arena findings enter the same taxonomy carrying their product-or-process tag, which is what lets one readout speak to the roadmap and the sales motion at once.
2. Quarterly readout, cross-functional by design: sales sees process failures, product sees capability gaps ranked by revenue impact, marketing sees the language buyers actually used, leadership sees the trend lines. One page of findings, each anchored to quotes.
3. Route the outputs to their consumers: loss patterns into the revival library, recurring objections into enablement and proposal pre-handling, competitive findings into battlecards, and systematic causes into qualification gates (a cause that recurs five times is not a loss reason, it is a missing gate).

What Good Looks Like

The strongest programs are boring on cadence and uncomfortable in content: every decided deal mined within a week, interviews running on their sample without a launch decision each time, and a quarterly readout where at least one finding contradicts what the company believes about itself. The common failure is the opposite shape: a burst of post-mortems after a bad quarter, run by the reps on their own deals, producing reasons everyone already agreed with, then silence until the next bad quarter. You know it works when the CRM loss field stops being quoted in decisions because a better source exists, and when a positioning or roadmap choice can cite the interview evidence behind it. And by the second quarter the program audits itself: put the CRM loss fields and the interview findings for the same deals side by side; the divergence table is your local proof that the method surfaces what the fields miss, and if the divergence is small, examine the interviewing before concluding your fields were honest all along.

Diagnostic Questions

1. Pull your last twenty losses. How many distinct loss reasons does the CRM show, and would anyone bet a euro on them?
2. When did a buyer last tell your company, in their own words and after the decision, why you really lost? Who heard it, and where is it written?
3. Which of your last quarter's losses were interviewed by someone other than the rep who lost them?
4. Name one product, positioning, or process change in the last year that cites decided-deal evidence. If none exists, where do those decisions currently get their facts?
5. Do your win stories and your marketing claims name the same differentiators? If you have never interviewed wins, how would you know?

Provenance notes and the practice-based rules list: read references/win-loss-provenance.md.

Built by Neon Triforce