Highlights

Duplicates Finder now records manual cleanup changes

Duplicates Finder already produced a Changes File for automatic merges. But when you opened a duplicate group and edited a row, deleted a row, or merged records manually, those changes could be missing from the file.

The Changes File now records those manual operations too, with before-and-after values for edited rows and the destination row for merged records. If you clean a CRM list by hand, you can download a complete log and use it to reproduce the changes in your CRM.

Improvements

Confidence scores for company and LinkedIn page matching

The Company Domain/Website and LinkedIn Company Page Matcher now returns a Confidence output when you start from a company domain or website.

A result is marked high confidence when the website listed on the matched LinkedIn company page agrees with the domain you supplied; otherwise it is marked low confidence. You can accept strong matches automatically and send uncertain ones to review instead of treating every returned page as equally reliable.

LinkedIn Search Scraper can add verified work emails

The LinkedIn Search Scraper can now look for a verified work email while it imports people from a LinkedIn or Sales Navigator search.

Previously, the source returned profile and company fields only. You can now enable Include Emails and map the email output into your collection—for example, turn a Sales Navigator search for French SaaS founders into a lead list with profile data, company data, and verified work emails in one run. Email lookup remains optional and uses additional credits only when enabled.

More reliable column management and CSV imports

Reordering columns from Manage properties now behaves predictably: drag a column to its new position and the table keeps that order.

We also fixed CSV headers that could not be used safely as Datablist property identifiers. A column such as 2026 Revenue now imports correctly while keeping its original name in the table, instead of creating a broken column.

Stoppable enrichment runs

Clicking Stop could end the current request while follow-up batches from a multi-step source continued to be queued. Supported sources now check the stop request before scheduling their next batch.

For example, if a source is processing a long result set in several batches, stopping the run prevents the remaining batches from being added while keeping the rows already imported.

See how to merge duplicate leads without losing data.