RevOps All-Stars
Home / Guides

A HubSpot CRM cleanup checklist for the first two weeks

Merges can't be undone and deletes only partly can, so both wait until the groundwork is done.

By the RevOps All-Stars team · 19 November 2026

HubSpot can't unmerge two records, and its recycle bin only gives you 90 days to bring back a contact you deleted. So the order you clean a portal in matters more than the list of things to clean, and we'd leave duplicates until the second week.

We read the pages ranking for this query. They agree on the items (duplicates, unused properties, lifecycle stages, bounced emails, old lists) and mostly treat them as a quarterly chore, as in the HubSpot community's quarterly checklist. One agency page stretches it into a 90-day program with a data council. Neither fits a new admin who inherited a portal they didn't build and has two weeks to show the sales team some progress.

Days 1 to 3: change nothing, find the leaks

Start with an export of contacts, companies and deals, with every property. It's your only real undo. HubSpot's merge docs say flatly that it's not possible to unmerge records, and the restore tool brings back deleted records for up to 90 days, 10,000 at a time, with the warning that some associated data will still be lost. Permanent and GDPR deletions skip the recycle bin entirely.

Then find out what's writing to the CRM. Open every active workflow and note which ones set or clear properties, especially lifecycle stage and owner. Do the same for integrations and imports, including the spreadsheet someone quietly re-imports every Friday. In our experience, most duplicates trace back to one or two of these. A commenter on that community checklist says the form or integration creating duplicates is usually the real issue, and it's the best line on the page. Merge before you plug the leak and you'll be merging the same people again next month.

Last, write down a baseline: total contacts, how many have no owner, how many have no lifecycle stage and how many open deals have no close date. Keep it to one page. In week two they're how you show the sales leader that something changed.

Days 4 and 5: owners and lifecycle stages, before any merge

In our view this is the step that decides whether merges go well. When HubSpot merges two contacts, it keeps the lifecycle stage that sits furthest down the funnel, per the same merge docs. A lead merged with a stale record marked Customer comes out as a Customer. If your stages don't mean anything yet, every merge spreads the confusion further.

So agree on the stage definitions first, in writing, with whoever runs sales and marketing. One sentence per stage, describing what has to be true for a record to be in it. Then find the workflows that set stages and check they match what you wrote. We'd bet a fair bit there's an old one in there still marking every form fill as an MQL.

Owners come next. Pull every record owned by a deactivated user and reassign it, by territory or to one named person if there's no rule yet. Check any rotation pools too, because a rep who left three months ago may still be sitting in one. Our guide to round robin routing in HubSpot covers how those pools fail quietly.

Week two: duplicates, then the contacts you're paying for

With that settled, start merging. Two things to know about HubSpot's own tool first, both from its duplicate management docs.

The first is what plan you're on. The duplicates manager needs a Professional or Enterprise subscription and shows up to 10,000 suggested pairs, one at a time. Bulk merging needs Data Hub Professional or Enterprise, which also raise the number of pairs shown to 30,000 and 100,000 respectively. On Starter you're doing it by hand from lists and exports, or paying for a third-party tool.

The second is how it matches. Contacts are compared on name, email, phone, zip code, IP country and company name, and companies on domain, name, country, phone and industry. So the tool is good at the obvious pairs and misses the person who signed up with a Gmail address and later with their work email. Sort those by hand for your open pipeline and your customers, and leave the long tail for later.

Then work in this order. The merge rules below come from the merge records docs mentioned earlier.

  1. Merge companies before contacts, so the people you merge next already point at one clean account.
  2. Pick the primary record on purpose. HubSpot keeps the primary record's values and only fills blanks from the other one, so the primary should be the record with the open deal or the latest real activity.
  3. Do your customers and open deals by hand. Bulk-merge only the clear matches, and reject the pairs you're unsure of (the duplicates docs give you 14 days to undo a rejection).
  4. Keep your own log. Per the duplicates docs, exported merge history covers auto-merges and rejected suggestions, but not merges done manually.
  5. Watch for heavily merged records. A merge fails once the two records have 250 merges between them, which tends to happen to the one contact every sync keeps recreating. That record points straight at your leak.

Then deal with the contacts nobody emails. Email verification vendor Truelist suggests treating contacts with six months of no engagement, and no open deal, as candidates for deletion after a re-engagement attempt. We'd be slower about deleting. Setting them as non-marketing gets you most of the benefit and keeps the history, since HubSpot says non-marketing contacts don't count toward your marketing contact tier and you aren't billed for them. Two catches for whoever owns the budget. The change only takes effect after your next update date, and HubSpot keeps you in the higher tier until your subscription renews. Don't promise finance a saving this month.

Properties last, and lightly

Property cleanup feels productive and rarely matters in the first fortnight. HubSpot's data quality tools flag properties that are duplicates, have no data or go unused. The list refreshes daily and is a fine place to start. But before you archive anything, search your workflows and reports for it, lists too. A property with no data can still be the trigger for a workflow that fires twice a year.

For the first two weeks we'd hide properties and leave archiving alone. Hidden properties drop out of the counts in that same report, so you get a cleaner view while keeping a way back. Archive in month two, once nothing has broken.

While you're in there, fix the input side. Make the few fields your reports depend on required on the forms and record creation screens. Switch free-text fields that people filter on, like industry or lead source, to dropdowns. The data quality tool can fix formatting problems such as all-caps names on Starter plans, but fixing them in bulk needs Data Hub Starter, and automatic fixes for new records need Data Hub Professional.

When this shouldn't be a hire

A cleanup with a finish line doesn't need a full-time person. If your portal is small, your sales team is a handful of reps and the main problem is a backlog of duplicates, our read is that a contractor on a fixed fee, or someone on your team working from this list, will get it done. We compare the options in HubSpot admin or a HubSpot agency.

The cleanup is the easy part to budget for. The mess comes back when nobody owns the leaks you found on day two, like the Friday import or the workflow that keeps resetting stages. So the decision rule is about ownership. Name the person who'll check that list of leaks every week from now on, and a one-off project is enough. Without one, expect to be running this checklist again by summer.

Hiring a RevOps person?

RevOps All-Stars places vetted RevOps specialists, HubSpot and Salesforce admins full-time on your team from $4K a month, all-in. One call to meet them.

Let's talk