Data decays on its own
This is the part that surprises people: nobody has to do anything wrong for a CRM to become inaccurate. People change roles and employers, companies rebrand and get acquired, offices move, and direct dials are reassigned. Every one of those events silently invalidates a record that was correct when it was entered.
So the goal is not a clean database. It is a database whose inaccuracy is visible and bounded.
What goes stale first
| Field | Decays | Because |
|---|---|---|
| Job title | Fastest | Promotions and reorgs, with no external signal |
| Direct phone | Fast | Reassigned when people leave |
| Email address | Medium | Bounces tell you, eventually |
| Company name | Medium | Acquisitions and rebrands |
| Person's name | Slowest | Rarely changes |
Notice that the fastest-decaying field is also the one with no automatic signal. A dead email bounces; a wrong job title just sits there being wrong, and gets used in a merge field.
Capture less
The most reliable cause of bad data is a form with too many fields. Every optional field is either left empty — in which case it is clutter — or filled in badly to get past it, in which case it is worse than empty.
Ask of each field: has anyone made a decision using this in the last quarter? Fields that fail should be archived. Fewer fields, filled in properly, beat a complete-looking record that nobody trusts.
Capture automatically
Anything a person has to remember to record will eventually not be recorded — not through carelessness, but because logging an activity is work that produces no immediate benefit to the person doing it.
The durable fix is for the record to update as a side effect of the work: mail sent from a connected mailbox appearing on the contact without anyone copying it across, meetings landing on the record when they are booked, a lead captured in one click at the moment you find it. Every one of those removes a step where accuracy was previously optional.
A field that stays accurate only through discipline is a field that will not stay accurate.
The routine
- Weekly, 10 minutes: close out deals with no next action and no contact in twice your cycle length. This one keeps the forecast honest and is worth more than all the others combined.
- Monthly: work the bounce list. A hard bounce is a confirmed dead address — mark it, do not re-send to it.
- Quarterly: run a duplicate check on email address, then on company name. Merge, do not delete, so history survives.
- Annually: archive contacts with no activity in two years. Archive rather than delete — reversible, and out of the way.
Delete without guilt
There is a strong instinct to keep every record on the grounds that it might be useful. It rarely is, and it has a cost: a contact list padded with dead addresses depresses every rate you measure, worsens deliverability when you mail it, and makes search slower to use.
In many jurisdictions there is a legal dimension too — holding personal data indefinitely with no purpose is hard to justify under data-protection law. Archiving what you have not touched in two years is defensible housekeeping in both directions.
For what a record should hold in the first place, see what is a CRM.