12 KPIs RevOps qui prédisent vraiment le revenu (et 8 à arrêter de suivre)
Les métriques RevOps qui corrèlent avec de vrais mouvements de pipeline dans Salesforce, les KPIs vanity que vous pouvez enlever, et les rapports qui les rendent visibles à la direction.
Most mid-market RevOps dashboards track 30 to 50 metrics. Fewer than 12 actually predict revenue. The rest are activity noise or lagging indicators the board notices too late. Here are the 12 KPIs we instrument into every Salesforce org we build, and the 8 vanity metrics we retire on day one.
The 12 KPIs that predict revenue
Top of funnel (leading)
- Marketing-sourced pipeline coverage: pipeline value in-stage divided by target. Watch weekly.
- MQL-to-SQL conversion: the shared handoff metric between marketing and sales.
- Speed-to-lead: time from form submission to first meaningful outreach. Target under 5 minutes for hot inbound.
Middle of funnel (predictive)
- Stage conversion rates: not a single win rate, a per-stage rate. Where deals actually leak.
- Cycle time by stage: aging in stage is the earliest sign of a stalled deal.
- Pipeline coverage: open pipeline value divided by remaining quota. Target 3x by quarter start.
- Weighted forecast vs plan: the RevOps forecast, not the rep-call forecast.
Bottom of funnel (accountability)
- Forecast accuracy: commit vs actual, tracked over 4 rolling quarters.
- Win rate on qualified opps: excludes junk, isolates real sales performance.
- Average sales cycle: not the average deal size, the average time to close.
Post-sale (predictive of retention)
- Net Revenue Retention: the single most important number in SaaS.
- Time to first value: from close to first customer outcome, a strong churn predictor.
The 8 vanity KPIs you can retire
- Total activities logged. Noise, not outcomes.
- Number of emails sent. Reply and meeting rates matter, volume does not.
- Number of leads (without lead quality or conversion). Meaningless in isolation.
- Cost per lead (without cost per opportunity or close). Optimizes the wrong thing.
- Average deal size (without win rate). Big deals do not matter if you lose them.
- Social media impressions. Not a revenue KPI.
- Pipeline created month-to-date without pipeline coverage. Numerator without denominator.
- Rep-call forecast alone. Should always be paired with a systemic forecast.
How to instrument these in Salesforce
| KPI | Salesforce object | Best surfacing |
|---|---|---|
| Pipeline coverage | Opportunity | Executive dashboard, weekly snapshot |
| Stage conversion | Opportunity History | Sales Cloud Funnel report |
| Cycle time by stage | Opportunity History | Custom report with time-in-stage formula |
| Forecast accuracy | Forecasts | Einstein Forecasting comparison |
| Speed-to-lead | Lead + Task | Rollup formula on Lead |
| NRR | Contract + Opportunity | Data Cloud calculated insight or CRM Analytics |
The KPI cadence that actually drives behavior
Metrics without cadence become wallpaper. In every RevOps engagement we install four cadences: weekly rep pipeline review (per-rep coverage), weekly forecast call (variance to commit), monthly funnel review (stage conversion vs target), quarterly business review (NRR, CAC, cycle time trend). Read our executive dashboards guide for exact report layouts.
Frequently asked questions
How many KPIs is too many?
An executive dashboard should have between 6 and 10 metrics. Rep-level dashboards, 4 to 6. Anything more becomes noise and hides the signal.
Should marketing and sales share the same KPIs?
They should share the funnel-conversion KPIs (MQL to SQL to SAL), and be measured on their own leading indicators separately. Shared metrics kill blame games.
What if we do not have historical data to set targets?
Use industry benchmarks for one quarter, then rebase from your own data. Do not skip target-setting because data is imperfect.
How often should we retire KPIs?
Every quarter. A KPI that no one has acted on in 90 days is decoration.
Si ça ressemble à votre CRM, regardons-le ensemble.
Trente minutes, sans slides, sans pitch commercial. Vous repartez avec un diagnostic dans tous les cas.