A Data Hygiene Schedule for Small Teams: What to Check Weekly, Monthly and Quarterly

The Problem With “Clean It When It Gets Bad”

Data does not get bad all at once. It drifts: a new form adds a field nobody maps, a rep starts typing countries differently, an import brings in 400 duplicates, a tool changes its export format. Each change is small and none is noticed until a report is wrong in front of someone who matters. The fix is not a bigger cleanup; it is a short, boring schedule that catches drift while it is still small. This one is designed for a team with no data engineer and about an hour a month to spend.

Weekly: 10 Minutes, One Person, Same Day Every Week

• New duplicates. Export contacts created in the last 7 days, run them against the full list by email. In a team adding 200 contacts a week, 5 to 15 duplicates is normal; 50 means a form or import is misbehaving. The Duplicate Remover does this in your browser.
• Imports that happened. Who imported what, from where, how many rows. One line in a shared note. The question in six months will be “where did these 3,000 contacts come from”, and this is the only place the answer will exist.
• Anything that failed. Bounced emails above your normal rate, a sync error from an integration, a form that stopped submitting. Note it; fix it if it is five minutes, schedule it if not.

That is the whole weekly check. It works because it is short enough to actually happen.

Monthly: 30 Minutes on the First Business Day

• Health score. Export the full contact list and run it through the Data Health Score. Write the number down next to last month’s. A drop of more than 3 to 5 points means something changed; find it before it compounds.
• Completeness on reporting fields. The five fields your dashboards depend on: what percentage is filled? A field that was 85 percent complete and is now 70 percent has a leak somewhere, usually a new source that does not capture it.
• Dead records. Deals with no activity past your threshold (60, 90 or 180 days, whichever you decided) get closed. Contacts owned by anyone who left get reassigned.
• Value drift. Pull the distinct values of your controlled fields (country, industry, lead source). Any new spelling that is not on the list gets mapped and the list gets updated. Ten new variants in a month is normal; a hundred means someone is typing into a field that should be a dropdown.
• Validation sample. Pick 100 random contacts and check that the email, phone and company fields look right. Not every record, just enough to notice if a whole category is wrong.

Quarterly: Two Hours With the People Who Use the Data

• Review the rules. The merge rule, the dead-deal threshold, the required fields, the controlled lists. Do they still match how the team works? A rule nobody follows is worse than no rule, because it creates the illusion of a standard.
• Full dedupe including near matches. Monthly catches exact duplicates; near duplicates (name and company match, different email) accumulate and need a human pass once a quarter. The merge guide covers the sort-then-dedupe pattern.
• Email list validation. Run the full list through a deliverability check before the next big campaign, not after the bounce rate spikes. See the deliverability checklist.
• Retire sources that only bring junk. If one lead source has produced 500 contacts and two conversations in a year, stop importing from it. Fewer bad rows beats more rows.
• Check what you are keeping and why. Contacts with no activity in two years and no legal reason to retain are a liability under most privacy regimes and a cost in every tool that bills per record. Archive or delete under a written policy.

The One Number

If the team will track exactly one thing, make it the monthly health score. It rolls completeness, validity, duplicate rate and consistency into a single figure, it is comparable month to month, and a sudden drop is the earliest warning you will get. Plot it on a line. When a new tool, a new rep or a new campaign changes the slope, you will know which.

A Realistic Example

A six-person recruiting agency with 12,000 candidate records and 1,800 client contacts adopted this schedule in April. Weekly checks found an ATS integration that was creating a duplicate candidate on every status change; fixed in week two, after it had already produced 340 duplicates. The monthly score went from 71 to 84 over four months, mostly from fixing phone formats and filling missing locations. The quarterly review retired one job board that had produced 900 candidates and no placements. Total time: about 90 minutes a month. The alternative, which they had done the previous two Decembers, was a two-day cleanup that was out of date by February.

When to Hand It Off

The schedule assumes someone on the team owns it. When that person leaves, or when the list passes a size where the monthly pass takes a day rather than 30 minutes, the choice is to hire for it or to outsource the recurring part. That is what the Data Care Plan covers: the monthly pass, the report, and the controlled-list maintenance, with the weekly check staying in-house because it needs someone who knows what imports are supposed to happen. Whichever way you go, the schedule above is the specification; a provider who cannot describe their version of it is not maintaining your data, only cleaning it occasionally.

Frequently Asked Questions

How often should a small team clean its CRM data?

A 10-minute weekly check for new duplicates and imports, a 30-minute monthly pass covering health score, completeness, dead records and value drift, and a two-hour quarterly review of rules, near-duplicates and sources. The schedule matters more than the depth.

What is a good data health score to aim for?

The absolute number matters less than the trend. Track it monthly; a drop of more than 3 to 5 points in a month means something upstream changed and should be found. Teams that start in the 60s typically reach the 80s within a few months of a regular schedule.

Should I delete old contacts?

Contacts with no activity in two years and no legal reason to retain cost money in per-record tools and are a liability under most privacy rules. Archive or delete them under a written retention policy, decided at the quarterly review rather than ad hoc.