Cleaning up a messy HubSpot CRM comes down to five moves, in this order: audit what exists, deduplicate records, prune unused properties, repair lifecycle stages, and retire the automations nobody remembers building. Do them in that order and nothing breaks. Do them in reverse — delete first, understand later — and you will spend a quarter explaining why the renewal report went blank.
Most messy portals aren't the result of laziness. They're the result of turnover. Three admins ago, someone had a system. Two admins ago, someone had a different system. Now you have 400 properties, duplicate contacts arguing with each other, and workflows that fire for reasons lost to history. Here's how to dig out.
Why should you audit before deleting anything?
Because in HubSpot, everything is connected to everything — and deletions are quiet until they aren't. A property that looks abandoned might feed a lead-routing workflow. A list that hasn't been opened in a year might power a suppression rule keeping unsubscribed contacts out of your emails.
Your audit should produce four inventories before you touch a single record:
- Properties — export the full list per object, note which have values on fewer than a handful of records, and check each one's dependencies (workflows, lists, reports) before flagging it.
- Workflows — every active workflow, who built it, and what it actually does. If nobody can explain one, it goes on the archaeology list, not the delete list.
- Lists — active vs. static, last-used dates, and which ones gate email sends.
- Integrations — every connected app writing data into the portal. Half your "mystery data" usually traces back to one of these.
This takes a few days of honest work. It saves months of forensic regret.
How do you deduplicate HubSpot contacts and companies?
Start with HubSpot's built-in duplicate management tool, but don't trust it blindly — review before merging, because merges in HubSpot are permanent. There is no unmerge button. That single fact should shape your entire dedupe strategy.
Work in this sequence:
- Fix the source first. If a form, import process, or integration keeps creating duplicates, deduping is bailing a boat with a hole in it. Patch the hole.
- Merge obvious pairs — same email, same person, clearly one human. HubSpot's tool surfaces these well.
- Handle the judgment calls manually. Two contacts named J. Smith at the same company might be one person or two. When in doubt, don't merge — flag for the record owner to confirm.
- Companies last. Company duplicates often carry different associated deals and tickets. Merge carefully and verify the associations landed where you expect.
For portals with thousands of duplicates, a bulk approach with tight matching rules beats heroic manual effort — but the no-unmerge rule still applies. Conservative beats fast.
Which properties should you prune — and which should you keep?
Prune properties that are empty, redundant, or unexplainable; keep anything with active dependencies or historical reporting value. A good rule of thumb: if a property has values on almost no records, no dependencies, and no one on the team can say what it's for, archive it.
Two cautions worth their weight:
- Redundant isn't always deletable. Three fields that all seem to store "industry" may be fed by three different integrations. Consolidate the data into one field first, then retire the others.
- Historical properties earn their keep. A field nobody edits anymore might still power a year-over-year report someone runs every January. Check report dependencies, not just workflow dependencies.
Deleting a property deletes its data on every record. If you're even slightly unsure, export the values first. Storage is cheap; reconstructed data is not.
How do you repair broken lifecycle stages?
Lifecycle repair means defining what each stage should mean, fixing the records that violate that definition, and then locking down what's allowed to set stages going forward. In most messy portals, lifecycle stage has been set by six different workflows, two imports, and a few well-meaning humans — so the field means nothing and every funnel report built on it is fiction.
Agree on stage definitions with both marketing and sales in the same room. Then run filtered views to find the violations: customers sitting at lead, MQLs with closed-won deals, opportunities with no deal attached. Fix them in bulk where the logic is clean, manually where it isn't. Finally, reduce the number of things that can write to the field — ideally a small set of documented workflows and nothing else.
What is workflow archaeology, and how do you turn off zombie automations safely?
Workflow archaeology is the practice of investigating old automations before deactivating them — because a workflow you don't understand is a workflow you can't safely kill. The zombie workflow that seems useless might be the only thing assigning leads in one region.
The safe shutdown pattern: read the workflow's enrollment history to see if it still fires, trace what it modifies, rename it with a clear review tag, then turn it off and watch for a couple of weeks before deleting. If something downstream breaks, you can reactivate in seconds. If nothing does, delete with confidence. Deactivate-and-wait costs you nothing; delete-and-pray costs you weekends.
How do you keep the CRM clean after the cleanup?
Clean CRMs stay clean through structure, not discipline. Nobody wins a long-term fight against entropy with good intentions — you win it with permissions, validation, and a recurring calendar invite.
Restrict who can create properties. Turn on validation rules for the fields that matter. Make forms and integrations conform to your picklists instead of writing free text. And book a quarterly hygiene review: duplicates, new properties, workflow inventory. An hour a quarter keeps you off this page forever.
If your portal's mess is deeper than a quarter's worth of cleanup — or nobody internal has the time to do the archaeology — this is exactly the kind of work our HubSpot team does for clients. We're a Platinum Solutions Partner, and untangling inherited portals is half the job. Talk to us before you delete anything you'll miss.
Frequently Asked Questions
How do I find duplicate contacts in HubSpot?
HubSpot's built-in duplicate management tool (under Data Quality) surfaces likely duplicate contacts and companies by comparing names, emails, and domains. Review each suggested pair before merging — HubSpot merges are permanent and cannot be undone — and fix the source of duplicates, like a misconfigured form or import, so they stop recurring.
Can you undo a merge in HubSpot?
No. Merging contacts or companies in HubSpot is permanent — there is no unmerge feature. The merged record keeps both activity histories, but you cannot split them apart again. That's why any CRM cleanup should merge conservatively: confirm two records are truly the same person or company before combining them.
How long does a HubSpot CRM cleanup take?
A focused cleanup of a mid-size portal typically takes several weeks: a few days for the audit, then deduplication, property pruning, lifecycle repair, and workflow review in sequence. Severely neglected portals with heavy automation debt take longer. The audit phase determines the real scope — never skip it to save time.