Cleaning Up a Messy Salesforce Org: A Partner's Checklist

The question behind the question

"Should we clean up our Salesforce or start over?" is almost never the right first question. The right first question is: do we actually know what we have?

Most firms don't. They inherited an org from a departed admin, worked with three different consultants over four years, and can't confidently say what half the fields, Flows, or automations actually do. Before anyone can decide whether to clean up or rebuild, someone needs to take a disciplined inventory.

Here's the checklist I run through with every firm that brings me an existing org.

1. What percentage of custom fields are actually being written to?

Run a report of field usage over the last 90 days. If more than 40% of custom fields on your top three objects haven't been touched, you have dead weight — but dead weight is cheap to remove. It's not, by itself, a reason to rebuild.

2. How many Flows, Process Builders, and Workflow Rules do you have?

Count them. Then count how many fire on each of your top three objects. If the same object has more than five Flows and multiple older automations, you almost certainly have conflicts or redundancies. Consolidation is usually possible — rebuild isn't necessary.

3. Is your core data model still correct?

This is the one question where the answer can force a rebuild. If the original implementation modeled Matters as Opportunities, or if Accounts are being used for both firms and prospects without separation, your foundation is wrong. Fixing a wrong data model in place is usually more expensive than starting over.

4. Are your users working around Salesforce?

Are there shared spreadsheets "just to be sure"? Email threads that duplicate what should be in the system? A second tool someone bought because Salesforce "doesn't work for us"? Those workarounds are the strongest signal that something's wrong — and they often reveal exactly what needs fixing.

5. Can a new user be productive in under an hour?

Sit a new attorney in front of the system. Ask them to open a matter and update its status. Time it. If it takes more than 10 minutes and requires someone to walk them through it, your page layouts and process are broken. Fixable — not rebuild territory.

6. Do reports match reality?

Pick the one report your managing partner looks at most. Ask three people what number it should show. If they disagree, the system doesn't have a shared definition of reality. That's almost always a data-hygiene and field-definition problem, not a platform problem.

7. Is there documentation?

If nobody can hand a new consultant a doc explaining the data model, the automation, and the known gotchas — you're in "tribal knowledge" territory. This isn't cause to rebuild, but whatever cleanup you do, documentation has to be part of the deliverable.

8. How old is the oldest Flow/automation, and does anyone know what it does?

Automation that nobody can explain is risk. The first step of any cleanup is building an automation map. If the map is scary, you have a rescue project. If it's merely ugly, you have a cleanup project. Those are different quotes.

9. What would you be losing if you rebuilt?

Real data — years of history, client interactions, matter records — has value. If you can preserve that by cleaning up in place, you avoid a painful migration. If the data is already broken beyond repair (duplicates, inconsistencies, junk), sometimes a rebuild with a clean data import is actually less work.

How to decide

After running through those nine questions, you'll be in one of three camps:

Cleanup (most common). Core data model is sound, but there's technical debt — dead fields, redundant automation, missing documentation. Scope a 3–6 week cleanup project. This is 70% of what I see.

Cleanup plus refactor. Data model is mostly right but has one or two structural mistakes (wrong object relationship, misused standard object). Scope 6–10 weeks — longer than straight cleanup, still shorter than rebuild.

Rebuild. Data model is fundamentally wrong, or cleanup would cost more than starting fresh. Rare, but real. In that case, treat it as a full implementation and bring over only the data that's worth preserving.

The cheapest cleanup

The cheapest cleanup is the one that includes documentation. Without documentation, you're guaranteed to be back in the same state in 18 months. With it, the next admin or consultant picks up where you left off instead of starting over.

If you'd like a second opinion on your org, book a 30-minute audit call. We'll tell you straight which of those three camps you're in.