HubSpot's duplicate tool is built around exact matches. If two contacts share an email address, or two companies share a domain, it finds them. Most of the duplicates I see in real accounts don't look like that.
The ones that slip through
Nicknames. "Bob Miller" from a webinar signup and "Robert Miller" from a sales import, same company, different email addresses. Nothing exact matches.
Phone formats. (555) 123-4567, 555.123.4567, +1 555 123 4567 and 5551234567 are one number. Stored as text, they're four.
Company suffixes. Acme, Acme Inc, Acme, Inc. and The Acme Company. Add www.acme.com on one record and acme.com on another, and the domain match fails too.
Gmail tricks. j.smith@gmail.com and jsmith+events@gmail.com go to the same inbox.
Finding them by hand
You can get a long way with exports and a spreadsheet:
- Export contacts with first name, last name, email, phone and company.
- Add a column that strips phones down to digits (and drops a leading 1 for US numbers).
- Add a column that lowercases company names and removes inc, llc, ltd, co, corp and a leading "the".
- Sort by the cleaned columns and look for neighbors.
- For names, keep a short nickname list (Bob/Robert, Bill/William, Liz/Elizabeth, Kate/Katherine) and check last name plus company.
Two rules worth keeping whatever you use:
- Never merge on name alone. Two John Smiths at different companies are two people. Look for a second signal: same company, same work domain, same phone.
- Watch shared phones. Couples and coworkers often share a number. A matching phone with a different first name is a household, not a duplicate.
And before you merge anything, export the losing record. HubSpot merges can't be undone.
The tool I built for it
I got tired of doing the spreadsheet version, so I built Dupe Sweep. It connects to HubSpot, scores every possible match, and lets you bulk merge the confident ones while the borderline ones wait for review. It works on Free and Starter accounts. HubSpot's own bulk merge needs Data Hub Professional.
I tested it against HubSpot's native duplicate detection on seeded test accounts with known duplicates and look-alike non-duplicates (households sharing a phone, coworkers, Ann and Anna). Native listed about three times more contact pairs than actually existed. Dupe Sweep's ready-to-merge group had no wrong pairs and, together with the review group, found about 98% of the real ones. On companies, native caught about 78% of strong duplicates because suffix and typo variants got past it.
There's a free plan with unlimited scans and 25 merges a month. If you try it, I'd like to hear what it missed.
Top comments (0)