Give Every CRM Field a Care Label or Retire It

Open your CRM and pick a field you did not create yourself. Industry, maybe, or Lead temperature, or a custom field called Status 2 that someone added for a campaign a couple of years ago. Now answer two questions about it without asking anyone: who is supposed to fill it in, and with which values?

If you cannot answer, you have found the real reason CRM data goes bad. The values are a symptom. Every field is a small promise that somebody will keep it current, and the fields nobody owns quietly break that promise every week. A cleanup fixes the values. Six months later the same fields have drifted back, because the promise still has nobody's name on it.

This matters more now than it did a few years ago. A salesperson knows to ignore Status 2. An AI feature that scores leads or builds segments does not, because it reads every field it is given as if it meant something. The free guide to AI CRM data hygiene covers the audit and cleanup side in detail. This post is about the step that keeps a cleanup from undoing itself: give every field a care label, and retire the fields that cannot get one.

What goes on a care label

Clothes come with a small label that says how to look after them, and nobody argues with it. A field needs the same thing, and it fits on five lines.

• Owner. A role, not a person, so the label survives staff changes. "Whoever creates the record" is a valid owner. "Marketing" is not specific enough.
• Filled when. The event that sets the value: the first call, the moment a quote is sent, the day a deal is won, the first invoice.
• Allowed values. A short pick-list wherever possible. Free text is where Tech, Technology and Software turn into three different industries.
• Stale after. How long a value stays true. A company's industry lasts for years. A lead's temperature lasts for days.
• Who may overwrite. A person, an import, an enrichment tool, or nobody once it is set.

That last line prevents a specific, common kind of damage. Enrichment tools and imports are useful, and they are also the fastest way to wipe out something a salesperson learned on the phone. "Enrichment never overwrites a value a person typed" is one of the most useful rules you can write, and it only exists if somebody writes it down.

One label, written out

Here is a finished label for a field most CRMs have and most teams neglect.

• Field: Lead source.
• Owner: whoever creates the record.
• Filled when: at creation, before the record is saved.
• Allowed values: Referral, Google search, Google Business Profile, Instagram, Website form, Event, Other (with a note saying what).
• Stale after: never. How someone first found you does not change.
• Who may overwrite: nobody after the first week. Imports skip the field whenever it already has a value.

Notice how much that settles. Two people can no longer record the same kind of lead in two different ways, and a later import cannot quietly turn a referral into a website form. It also forces a decision the team has probably been avoiding: what exactly counts as a referral? Settle it once on the label instead of in every Monday meeting.

Retire the fields that cannot get a label

Go through the field list one field at a time and try to write the five lines. Most fields will get a label in a minute. Some will produce a long pause and then a guess about who created it, for which campaign. Those are the ones to retire.

Retiring feels risky, so do it in two steps. First, hide the field from every form and view for a month, and keep the data. If nobody asks where it went, archive it, and export its contents to a file before you do. You lose nothing, and the CRM gets one field quieter.

I will say the unpopular part plainly. For most small teams, the cleanest CRM has fewer fields than the one they have now. Every field shown on a form is a question someone has to answer while a customer is waiting, and a field that is optional and unowned mostly collects guesses. Guesses are worse than blanks, because a blank at least admits it does not know.

Put the label where people type

A care label in a shared document helps whoever wrote it. The people entering data need it at the moment they enter the value. If your CRM has a description or help-text box for each field, paste a short version of the label there, so the rule appears next to the field itself. If it does not, a one-page sheet pinned where the team works is the next best thing.

Pick-lists do half the work on their own. Converting a free-text field to a short list of allowed values ends a whole class of problems, and the old free-text values are a good preview of what the list should contain. Group them, name the groups, and map every old value to one of them in a single pass.

The weekly and monthly chores

Care labels also tell you what maintenance the CRM needs, because each field's "stale after" line becomes a chore. In practice that turns into a short routine.

• Weekly: merge the duplicates created since last week, before they collect separate histories. The post on deduplicating contacts without losing history covers how to merge without losing notes and activity.
• Weekly: look at records created with an owner field left blank, and assign them.
• Monthly: filter each short-lived field for values older than its label allows, and recheck or clear them.
• Quarterly: reread every label. Retire the fields whose owner no longer exists, and add labels for anything new.

None of this needs software beyond the CRM you already have. It needs a name next to each chore, which is the same lesson as the label itself.

Before you connect an AI tool

If you are about to switch on AI lead scoring, AI segments or an assistant that reads your CRM, run the labels first. Then give the tool only the fields that have a label. I would rather give a model eight trustworthy fields than forty where a dozen are guesses, and the eight are far easier to explain when the tool gets something wrong.

The CRM data hygiene guide picks up from there. It includes a two-hour audit framework, a method for deduplication and merge at scale, a comparison of Clay, Apollo and ZoomInfo for enrichment, normalization with AI classification, entry-point quality gates and an ongoing maintenance system. It is a free PDF, and the care labels will make every chapter of it easier to apply.

Start with the field you picked at the top of this post. Write its five lines today, or hide it from the form.

← Back to all articles