Field Audits: Yearly Housecleaning for Your Salesforce Org
Every org starts clean. Someone builds the objects, adds the fields the business needs, and calls it done. Then a sales director wants one more field. Marketing wants three. Someone leaves the company, and nobody remembers what half those fields were for, only that deleting them feels risky. The org grows fields the way an old house grows junk drawers. One reasonable decision at a time, until you can’t find the scissors and everything is tangled in yarn.
Salesforce Field Usage: Why It’s Hard to Track
Salesforce doesn’t give you a built-in report showing what percentage of your records actually have a given field filled in. You’d think a data company would make that number easy to find. It doesn’t. Salesforce’s own help documentation confirms this and points admins to the AppExchange for a fix.
There are free, native options on AppExchange, and they’ve become something close to a rite of passage for org cleanup. Point one at an object, and it shows you, field by field, how many records actually have data in that spot. Run it against Contact or Account, and you’ll likely find fields sitting at two percent, three percent, five percent full. Fields nobody uses. Fields cluttering page layouts and adding confusion to every report that touches them. Read to the end for some examples of AppExchange tools you can use.
Why Field Descriptions Matter for Agentforce
Fill rate isn’t the only thing worth checking on a field. There’s a second, easier-to-miss problem: does the field have a real description attached to it? That’s a different question entirely from whether the field holds data. A field can be fully populated on every record and still carry a blank or lazy Description in Setup.
According to Salesforce’s own guidance, the Description field, the one most admins fill in halfheartedly if at all, now does double duty. It tells a human what a field is for, same as it always did. It also tells Agentforce the same thing. The agent reads that description to decide whether a piece of data is worth pulling into its answer. A field with no description, or a lazy one that just repeats the field’s name, gives the agent nothing to work with. A clear one tells the agent exactly what it’s looking at.
How to Start Cleaning Up Salesforce Fields
You don’t need to overhaul the whole org in a weekend, and you shouldn’t try. This approach follows a rotational audit: one object a month. Credit goes to Lizz Hellinga, who presented this method at Texas Dreamin’ this year.
- Pick one object: Start with whichever one gets touched the most, since that’s where the payoff is biggest. Accounts, Contacts, and Opportunities are the usual suspects.
- Run an AppExchange tool against it: Let the tool show you what’s populated and what’s gathering dust. Don’t guess. Guessing is how orgs end up full of assumptions.
- Sort the results into three piles: Fields worth keeping as is, fields worth keeping but needing a real description, and fields worth removing once you’ve confirmed no automation or integration quietly depends on them.
- Write descriptions a human would actually understand: Skip the jargon. Say what the field holds and why it exists, clearly enough that a new hire gets it on day one.
- Move to the next object next month: Consistency beats heroics here. A rotational habit clears your whole schema in a year. One frantic weekend doesn’t.
A year from now, your schema will look nothing like it does today, and every field left standing will have earned its place.
Duplicate Records: Why They’re a Data Quality Risk
Fields aren’t the only clutter working against you. Duplicate records cause the same damage from a different angle, and they’re often worse, since a duplicate doesn’t just sit there like an empty field. Three versions of the same Account, each with slightly different data, leave an agent guessing which one to trust, the same way a rep gets confused working a lead that another rep already closed out under a different record.
For example, say “Acme Manufacturing” shows up three separate times in the org:
- “Acme Manufacturing,” created by a sales rep from a cold call. Phone number only, no billing address, and the Account Owner is that rep.
- “Acme Mfg Inc.,” created six months later by a different rep working an inbound lead. Has a billing address and a website, but no phone number, and the Account Owner is the second rep.
- “Acme Manufacturing LLC,” imported through a data enrichment integration. Has the correct legal name and a D-U-N-S number, but the industry field is blank and it’s missing every open opportunity tied to the other two records.
None of these three records is wrong exactly. Each one just holds a different slice of the truth. A rep pulls up the first one and doesn’t see the open opportunity sitting on the third. An Agentforce action asked to summarize the relationship with Acme grounds itself against whichever record it happens to match, then gives someone a confidently incomplete answer, since it has no way of knowing the full picture is split across three places.
Salesforce doesn’t block duplicates on its own out of the box, whatever your org may assume. Admins have to build that protection themselves, deciding what counts as a match and what Salesforce should do once it finds one.
A rotational audit is a natural place to run a deduplication pass alongside your field review, since you’re already elbow-deep in the object anyway. Check whether your duplicate rules are active for the object you’re auditing this month, then let Salesforce’s own duplicate jobs flag what’s already sitting there unmerged. Merge what you can with confidence, and flag anything ambiguous for a second look rather than guessing. An org with clean fields and duplicate records everywhere hasn’t solved its data problem. It’s only solved half of it.
Continuous Improvement, One Object at a Time
Set a real target, not a vague one. Field profiling with a genuine goal of zero percent unnecessary fill rate forces a decision on every field you’ve been avoiding. Either it earns its place through real usage, or it goes. Nothing in an org stays neutral forever. It’s either helping or it’s hurting.
Nobody throws a party for cleaning up an Object Manager, but six months into a rotational audit, your reports get more trustworthy, your page layouts get shorter, and your AI features start pulling the right data more often. None of that happens by accident. It happens because someone decided the junk drawer was finally worth sorting through. Start with one object this month. The rest of the schema will still be there next month, waiting its turn.
For official Salesforce documentation click HERE.
Here are a couple of AppExchange tools to start with: FieldSpy, FieldPro
Explore related content:
AI Data Masking and Sensitivity Mapping in Salesforce
