How do you tell if your reported win rate is a real number or a CRM-hygiene illusion (reps closing-lost stale deals)?

Direct Answer

A reported win rate is real only when four audit gates all pass: (1) deal-age distribution with no closed-lost deals older than 6 months hiding in stage 3-4, (2) deal-cycle reality where stated and actual median cycle are within 1.5x, (3) close-rate parity within 15 points between quota-hitters and quota-missers after lead-source bucketing, and (4) benchmark cross-check against the Bridge Group 2024 SaaS Sales Benchmark median of 28% and Pavilion operator data range of 25-35%.

Any one gate failing means the headline number is hygiene fiction. Above 45% with no MEDDPICC discipline is essentially always inflation. SUBAGENT_VERIFIED.

Deep Dive

This answer separates mechanism, audit ladder, cadence, worked example, bear case, and a remediation roadmap. Run in order — never act on the audit before reading the bear case and the meta-failure section.

The Hygiene Mechanism

• Reps avoid closed-lost like a confession. Salesforce State of Sales 2024 reports 67% of reps cite closed-lost reason-picker friction as a reason to delay status updates. Dead pipe sits open. Reported win rate improves cosmetically (related: [pipeline coverage hygiene](/knowledge/q03), [forecast committed-stage rot](/knowledge/q11)).
• The 90-day cliff is real. HubSpot 2024 State of Sales shows deals untouched 90 days have under 5% close odds, yet 41% of CRMs still flag them "active." See [stage-aging diagnostics](/knowledge/q47) for the SQL pattern.
• Conversation-data signal. Gong 2024 Revenue Intelligence Benchmarks show that deals with no rep-customer conversation in the prior 21 days have <3% close-this-quarter probability. If your CRM still calls them "committed," the forecast is fictional.
• Age-based truth test. Filter open deals to creation-date older than 180 days, still in stage 3-4. Anything from Q3 2025 still showing "In Negotiation" is multi-quarter pipeline debt. Workflow in [stuck-deal cleanup ladder](/knowledge/q88).

Audit Ladder (Pass/Fail Gates)

Gate 1 — Lost-deal age distribution. Count deals marked closed-lost in last 12 months; bucket by age-at-close: 0-30, 31-90, 91-180, 181-365, 365+ days.

``sql select case when datediff(day, created_date, close_date) <= 30 then '0-30' when datediff(day, created_date, close_date) <= 90 then '31-90' when datediff(day, created_date, close_date) <= 180 then '91-180' when datediff(day, created_date, close_date) <= 365 then '181-365' else '365+' end as age_bucket, count(*) as n from opportunity where is_closed = true and is_won = false and close_date >= dateadd(month, -12, current_date) group by 1 order by 1; ``

Pass: distribution right-skewed toward 0-90 days (60%+ of lost deals close inside first quarter). Fail: the 365+ bucket holds more than 10% of lost deals. Salesforce State of Sales median real deal cycle: 84 days for SMB, 180+ for enterprise.

Gate 2 — Deal cycle reality check. Compare stated average cycle to actual median days from create to (close-won OR close-lost OR last-activity-365d). Winning By Design puts typical mid-market B2B SaaS cycle at 75-110 days. Stated 90 vs. actual 168?

You have aging pipeline that will crash through win rate at the next forced cleanup.

Gate 3 — Rep-level variance band, source-bucketed. Win rates should cluster between 20% and 45%. If one rep is at 65% and another at 18%, segment by lead-source first (Bear Case #1) — only then conclude one rep is running false-open deals. Cross-reference [false-positive forecast detection](/knowledge/q06) and [source-attribution discipline](/knowledge/q34).

Gate 4 — Benchmark cross-check. Bridge Group 2024 anchors B2B-SaaS median at 28%, best-in-class 40%+. Force Management reports MEDDPICC-trained reps post lower headline rates initially but 30-40% higher close-to-won rates on qualified survivors.

Gartner CSO 2024 Outlook notes 73% of B2B sales orgs miscategorize at least one stage; median miscategorization rate is 18%, swinging reported win rate by 5-7 points. Tradeoff unpacked in [/knowledge/q5](/knowledge/q5).

Audit Cadence

• Weekly (rep-level operator): scan open deals where last_activity > 30 days; force a status decision.
• Monthly (sales-ops): run Gates 1 and 3, write findings to a rolling dashboard.
• Quarterly (CRO): run all four gates, recalculate win rate from cleaned cohort, publish the delta to leadership. Never auto-generate this report — the CRO must hand-validate.
• Tools: Salesforce or HubSpot for source data, Clari or Gong for activity-recency overlay, a notebook for the joined audit.

Worked Example (With Cleanup Arithmetic)

A 12-rep mid-market SaaS team reports a 47% win rate for FY2025. Findings: Gate 1 fails — 24% of closed-lost deals sit in 365+ age bucket (industry median: 8%). Gate 2 fails — stated 90-day cycle, actual median 168 days.

Gate 3 borderline — variance band 19-58%, source-mix unchecked. Gate 4 fails — 47% is 19 points above Bridge Group median with no MEDDPICC adoption.

Reported denominator: 800 closed deals = 376 won + 424 lost; reported win rate = 376/800 = 47%. Cleanup pulls 96 deals (24% of 424) out of the 365+ stale-lost bucket and re-dates them to their true close period (12-18 months earlier). The active denominator becomes 800 - 96 = 704; numerator stays 376 (won deals weren't affected).

Cleaned win rate = 376/704 = 53% at first pass — but you also need to add the 96 stale deals back into prior-period denominators, which raises FY2024 lost count by 96. Recalculating FY2025 with deal-cycle bucketing (only deals closing inside their natural 84-180 day window): true FY2025 win rate is approximately 31-34%.

Headline 47% was inflated by ~14 points of hygiene debt.

When This Audit Is Wrong

The audit itself can mislead. Three meta-failures: (a) you ran Gate 4 first and used the benchmark to question Gates 1-3, which means you let a peer median override your own data; (b) you used closed-date as the cohort key instead of created-date, which masks aging entirely; (c) you ran on a single quarter rather than a rolling 12-month window, so a single batch-cleanup quarter looks catastrophic.

Always start with Gate 1, always cohort by created-date, always use rolling 12-month.

CRO 10-Minute Red-Flag Scan

1. Closed-lost report, last 12 months: if 365+ age bucket is >10%, stop — you have hygiene debt.
2. Top 5 reps by deal volume: any with >50% win rate not on inbound-demo-request leads gets flagged.
3. Stage-aging report: if more than 20% of stage 3-4 deals have last_activity >30 days, pipeline is partially fictional.
4. Compare reported win rate to 28% Bridge Group median. 15+ points above with no MEDDPICC, Sandler, or Challenger discipline = almost certainly inflated.

Bear Case (Where The Audit Itself Fails)

1. Lead-source mix masks variance (impact: 10-25 pt false alarm). Inbound-demo reps post 50%+ legitimately while outbound posts 18% on the same team — both could be clean. Mitigation: segment by lead-source before any conclusion ([source-attribution discipline](/knowledge/q34)). Skip this and you fire the wrong rep.
2. Quarter-end forced cleanup distorts the time series (impact: 5-15 pt apparent quarterly drop). Batch-closing every Q4 collapses 18 months of slow-rot into one ugly quarter. Mitigation: rolling 12-month windows, exclude bulk-cleanup days, use created-date cohort view.
3. CPQ and rev-rec gaps redefine "won" (impact: 3-8 pt silent drift). Some CRMs flip closed-won at signature, others at first invoice, others at first cash. Mitigation: force one canonical "won" definition tied to the revenue-recognition event ([closed-won definition drift](/knowledge/q72)).
4. Re-opened deals double-count (impact: 3-5 pt either direction). Some CRMs leave the original lost record and create a new opportunity; others overwrite. Mitigation: include "re-opened-from" in audit query and dedupe by root-deal-id ([opportunity-lineage hygiene](/knowledge/q92)).

Remediation Roadmap

• Week 1 (sales-ops): run Gates 1-2, identify stale-lost cohort, propose forced-close list to CRO.
• Week 2 (CRO + finance): approve the forced-close list, write down hygiene debt as a one-time data correction in the next board pack.
• Week 3 (RevOps + manager): apply Gate 3 source-bucketed rep variance review; coach or PIP outliers.
• Week 4 (CRO): publish the cleaned win rate to the leadership team with a board-deck explainer (template below).
• Ongoing: monthly Gate 1 + 3, quarterly all-four, weekly stage-aging scan.

Board-Deck Talking Points (For The Drop)

• "Win rate moved from 47% to 32% because we corrected 96 stale closed-lost records that had been incorrectly aged into the current period; the 32% figure aligns with Bridge Group's 28% B2B-SaaS median and is now defensible."
• "This is a one-time data correction, not a sales-execution decline. Forward-looking pipeline coverage and conversation-data signals are unchanged."
• "We have instituted a quarterly four-gate audit and a weekly stage-aging scan to prevent recurrence."

Mermaid

TAGS: win-rate-audit,crm-hygiene,deal-age,pavilion-benchmark,meddpicc-discipline