What is data hygiene?
Data hygiene is the ongoing practice of keeping records accurate, complete, de-duplicated and consistently formatted. Data hygiene is continuous maintenance rather than a one-off clean-up, because B2B records decay as people change jobs and companies change shape.
Why it matters
Every automated decision a go-to-market team makes — routing, scoring, segmentation, suppression — reads the database as though it were true. Bad records do not produce obviously bad output; they produce plausible output that is quietly wrong, which is harder to notice and more expensive to unwind.
What data hygiene covers
- Deduplication, so one company is not three accounts and one person is not two contacts.
- Standardisation of formats — country names, phone formats, job titles, industry values.
- Validation, so an address that no longer accepts mail is marked rather than mailed.
- Completeness checks on the fields your routing and scoring rules actually read.
- Decay management: re-verifying records on a schedule, and retiring the ones that fail.
How records decay, and what breaks
| Decay type | What changes | What it breaks |
|---|---|---|
| Job change | The contact is no longer at that company | Outreach lands at a former employer |
| Role change | Title and responsibilities move | Persona targeting and scoring |
| Company change | Headcount, name, ownership or address | Territory routing and qualification |
| Address change | A verified email stops accepting mail | Deliverability and sender reputation |
| Duplicate creation | The same account enters the system twice | Reporting, attribution, and double outreach |
Common mistakes
- Treating hygiene as a project. A cleaned database starts decaying the day the project ends.
- Deleting rather than marking. A suppressed record still needs to exist, so the same bad data cannot be re-imported.
- Standardising on import but not on the forms and integrations that write to the CRM afterwards.
- Measuring completeness across every field rather than across the fields your automations actually read.