Complete · green
Last changed 9 weeks agoRevenue decision guide · CRM evidence quality
CRM evidence quality: know what supports the field
Keep the CRM as the system of record while making the source, timing and uncertainty behind consequential fields inspectable.
The operating question
A populated field doesn't prove the underlying claim.
Every required field on the opportunity is populated. The data-quality report is green and the record passes the completeness check. It's also nine weeks since anybody spoke to the budget owner.
One field came from a Seller note, one was copied forward by an automation and another was written by an integration nobody has reviewed since it was enabled. All three look identical in the CRM. The record is complete, current in the technical sense and commercially stale.
This is why evidence quality isn't the same as field completeness. Completeness asks whether a value exists. A revenue decision needs to know who supplied it, what supports it, when it applied, when it was recorded and whether a person or system had authority to change it.
Applied date unknown
Write came from an integrationCompleteness measures typing
The obvious fix is a completeness score. It's useful for finding fields that are empty, broken mappings and required process data that never arrived.
It can't distinguish a current buyer-confirmed fact from an old assumption copied forward by a workflow. A full field tells you somebody or something answered. It doesn't tell you whether the answer can support the decision being made today.
A complete record is the easiest kind to be wrong about
Visible gaps invite questions. Green fields close them. That makes unsupported completeness more dangerous than an honest blank because the team stops looking.
Keep missing, stale and conflicting evidence as different states. Missing means no source arrived. Stale means a real source is older than the customer's policy allows. Conflicting means two attributable sources disagree. None should be silently converted into confidence for the roll-up.
Keep source attribution and two dates
For every consequential field, retain the source, owner and authority path. Keep when the fact applied separate from when it was recorded. A requirement agreed on 12 March and entered on 2 April should not be treated as first known on 2 April or as visible to a decision made on 20 March.
Freshness is separate from correctness. A fact can have been correct in March and still be unsafe for a September forecast. The customer should define how freshness works for each kind of evidence rather than applying one universal age.
Control the write, not only the read
A field changed by an integration or AI tool needs the same accountability as one changed by a person. Record the source value, proposed change, configured authority, approval where required and final write outcome.
The CRM remains the system of record. PROOF adds the decision record behind the fields that matter, and writes back only through the authority the customer has approved.
What the external evidence supports
Salesforce's State of Sales, seventh edition, reports that 46% of sales professionals whose teams use AI agents say data-quality issues hurt their sales. The report is based on an anonymous survey of 4,050 sales professionals across 22 countries, fielded from August to September 2025. Read Salesforce's report.
That finding concerns teams already using agents. It doesn't measure CRM evidence quality directly and it doesn't prove that source controls improve revenue. It supports the narrower point that AI increases the cost of weak commercial data.
Where evidence programmes go wrong
Don't require a source for every harmless field. Apply stronger evidence and approval rules where a value changes qualification, forecast, commercial commitment or customer treatment.
A commonly repeated Warwick Business School statistic remains excluded. The accessible material doesn't publish a denominator and the underlying report is gated. On a page about evidence quality, an unverifiable number would defeat the argument.
How PROOF approaches it
Keep the evidence, decision and action connected.
PROOF keeps the source, timing and accountable decision together so the team can act while the outcome can still change and learn from what happened afterwards.
See the working product