Executive search
Contact provenance in a CRM: keep the source attached to the useful fact
Distinguish CRM record origin from field evidence, effective dates, edits and transformations. Includes a practical provenance record and a synthetic four-field example.
A CRM says a contact came from an import. That explains how the row got there, not why a researcher should trust the role title on it. A work address may have come from the contact directly; a responsibility may be a consultant's inference. The fields need different checks before anyone uses them.
HubSpot's Record Source documentation describes how a record was originally created. Its property-history documentation describes changes to individual values, including the date and source of the change. Neither replaces evidence for the claim itself. If an enrichment job writes a title, the job explains the update; the underlying source still determines what the title establishes.
Give consequential fields their own evidence record
Start with the fields a consultant will carry into outreach, a shortlist or a client note. Leave low-impact fields alone. For a consequential field, capture:
| Record | What to enter |
|---|---|
| Value and entity | Exact claim, correct person and any assignment-specific context |
| Source and access | Link to the permitted source or identify the speaker; note who can see it |
| Date the claim describes | Start date, end date or period, if the source states one |
| Date checked | When someone inspected the source |
| How it got into the CRM | Entered directly, imported, enriched or inferred |
| Use and next check | Which output may use it, which use is on hold, and who can resolve the gap |
The date matters in two ways. If a company posts an appointment on 4 November effective 1 December, a check on 12 November tells you the team saw the announcement, not that the person already holds the new job. Preserve both dates instead of replacing the current title prematurely.
Worked example: one fictional contact, four evidence decisions
A team is preparing a market-map update and considering an introduction to Priya Rao. The following records are fictional.
| Field | Evidence and dates | Decision |
|---|---|---|
| Role title | Company notice published 4 November says “Head of Operations” effective 1 December; analyst checked 12 November | Label as announced, not current, in the November market map |
| Employer | Priya's 10 November reply says she is still with the same company | Use for the November map; revisit on the next evidence check |
| Work email | Confirmed in the reply, which asked to keep this conversation on the existing thread | Use the existing thread if follow-up is appropriate; do not treat the address as permission for broader outreach |
| Regional remit | Analyst inferred APAC responsibility from an event biography without a date | Leave the remit out of the client view; analyst to verify it with a source that names the scope |
The useful distinction is not whether the whole record passes. It is whether a value is usable for a specific output. An address may be accurate, for example, without being suitable for a shared export. Completeness and accuracy are separate dimensions in the UK Government Data Quality Framework; this field-level method is an adaptation for search teams.
Run one correction through to the output
Pick one consequential field and trace it from the source to the last output that used it. If a correction is needed, retain the earlier value as history where appropriate, change the current value, and check copies in candidate briefs or client notes. Log who fixed each copy. A practical pass condition: a second reviewer can identify the supporting source, say what date the fact describes, and tell you which output still needs correction without asking the original researcher.