How do I measure whether my lead scoring is working?
Check that high-scoring leads actually convert better, and adjust when they do not.
Measure whether lead scoring is working by comparing conversion rates across score tiers over time, so you can see that High Fit leads really do convert better, and adjust the model when the gap shrinks. A score that does not separate winners from losers is not doing its job.
This guide covers ongoing measurement once scoring is live. For the initial one-time check that a new model works before you trust it, see how do I validate a scoring model? Validation happens first, measurement never stops.
What is the core metric?
Conversion by tier. If High Fit leads convert meaningfully better than Moderate Fit, and Moderate Fit better than Low Fit, the score is predictive. If the tiers convert about the same, the model needs work, because the number is not telling reps anything they could not guess.
| Fit tier | What you expect | Read if it holds |
|---|---|---|
| High Fit | Highest conversion | Score is predictive at the top |
| Moderate Fit | Middle conversion | Tiers separate as intended |
| Low Fit | Low conversion | Right to deprioritize these |
| No Fit | Near zero | Right to filter these out |
What else should I watch?
Watch for drift. A score that predicted well six months ago can decay as your market, product, or ICP changes. Track the tier gap over time, not just once, and re-validate when it narrows, that narrowing is the early warning that the model is going stale.
Illustrative example
Illustrative, based on how SignalRx customers work. A company selling a care-coordination platform to home health agencies checks conversion by tier each quarter. High Fit leads, independent multi-branch agencies like a 7-branch group in Tennessee, convert at several times the rate of Low Fit. Two quarters later the gap narrows because a new competitor entered the mid-market, itself a news and web signal SignalRx reads. That is the signal to revisit the weights, not to abandon the score.
How does this feed back into the model?
When a tier stops predicting, revisit the signals and weights in validating a scoring model and adjust, and often the fix is refreshing the external signals behind the score, web, social, reviews, and registries, not just the weights. Measurement is what keeps scoring honest rather than decorative, and SignalRx writes the tier to Salesforce, which stays the system of record, so the conversion-by-tier report comes straight from your own pipeline. Under that report, fit for a not-yet-customer lead is built from external public data, provider taxonomy in NPPES, CCN-keyed CMS size and setting, and ownership signals like MSO or private-equity backing, while a customer's fit and health scores draw on your own internal signals of revenue, support, and usage over time. SignalRx scores accounts and contacts alike and lets inference decide which source to trust for each record, so a narrowing tier gap points you at the signal that has gone stale.
Key takeaways
- Measure conversion by score tier as the core test.
- Track the tier gap over time to catch drift.
- Feed the findings back into weights and signals.
See it in action: Close Better Leads with SignalRx.
Related: How do I validate a scoring model? and How do I keep Fit Scores current?
Want any of these running on your data?
Frequently Asked Questions
.svg)
.avif)