How a Fintech Gave Its Sales Team a Pipeline It Can Trust by Clearing 4,000+ Duplicate Matches with Dedupely

01

Where duplicates came from

One company, two website fields, and a deal for every duplicate

The company is a fintech business that runs its sales pipeline in HubSpot, with more than 500,000 records and a couple of thousand deals. The RevOps manager had used Dedupely a couple of years earlier, moved away from it, and came back with what he described as a big duplicate problem at the company level and the deal level. He had already built searches and merge rules of his own, and duplicates were still getting through.

Three things kept creating them:

• The website lived in two places. Some company records stored it in the domain field, others in a second website field. A search on one field missed the duplicates hiding in the other.
• Company merges created deal duplicates. When two companies merged, each brought its own deal along, so the surviving company ended up with two deals in the same pipeline. Every company cleanup could leave a new deal cleanup behind.
• Not every record should match. Companies created from personal email domains, like Gmail or Yahoo, and companies that belong to a parent company had to stay out of the merge, or the wrong records would combine.

02

What duplicates broke

A weekly hunt for records that should have merged

When he came back, he said the duplicates were causing chaos. Nearly every week he would find records that should have merged and hadn't, and wonder why there were still so many duplicates.

On the deal side, a merged company could end up with two deals for the same client in the same pipeline, and someone had to find and merge them by hand. The team had already decided which deal stages should survive a merge, with stages like Lost or Bad lead set to lose.

Why loose rules were risky

Matching on too little was risky at this scale. Early on, an automated company search merged two companies whose domains matched except for the ending. Domain Root ignores www, .com and country variants by design, so the search did what it was set up to do. He wasn't worried about that pair, but he paused the search's Auto Merge and added a second check, so two companies that really are different couldn't merge the same way.

Matching on too much missed real duplicates instead. A search that required both the domain and the website field skipped pairs where one record left the website blank, because blank values don't count as a match.

04

What deduplication changed

Searches that run themselves

Within a few weeks he was merging so much that he had stopped keeping count. Soon after, his main automated company search was doing its job, catching most duplicates and only occasionally flagging a real edge case, so he left it running. An older one-off search now returned nothing at all, most likely because its duplicates had already been merged.

The numbers back that up. In under 7 weeks, the team merged 2,158 company matches and 1,852 deal matches, many of them with three or more records in a single match. Twelve searches now run on Auto Merge, so new duplicates across companies and deals merge on a schedule instead of waiting for someone to spot them.

A tool the team kept building on

He described Dedupely as doing a great job deduplicating their records, and the team kept adding to the setup: separate searches for companies and deals, Merge Rules for each object, and filters that keep the wrong records out.