CRM Duplicate Cleanup

Every CRM rots the same way: the same customer gets entered twice. Bob Smith and Robert Smith. One with a phone number, one with a company. Sales works off whichever they happened to open, deals get double-counted, and nobody has an afternoon spare to sit and merge them by hand.

This flow does the merging. It finds the duplicate pairs the CRM has already grouped, and merges each one into a single clean record — the tedious housekeeping that never makes it to the top of anyone's list.

It runs against Zapspot, a fictional CRM used for training. The data is entirely synthetic.

What it produces

crm-cleanup.csv in your home folder — the before-and-after: how many duplicate pairs were merged and how many remain. On the seeded CRM that is 12 pairs merged, 0 left.

How it works

1. Open the duplicates view

Sign in and switch the contact list to its duplicates view. The CRM has already grouped the likely duplicate pairs — same person, name variants — so the robot doesn't have to guess who is who; it works through the pairs the CRM surfaces.

2. Merge every pair

Merging a pair removes it from the list, which would break a fixed loop counter. So instead the robot keeps merging the first remaining pair and re-checks, until there are none left. Each merge opens the side-by-side picker and confirms, keeping the primary record and folding the variant into it.

This loop-until-empty shape is worth remembering for any "work the list until it's clear" job where acting on an item removes it.

3. Report

The report is the proof: pairs merged, pairs remaining. Run it overnight and the CRM is clean by morning.

Running it

Ready to run as-is. It signs in with the published training credentials (jonas.weber@globex.example), which are not secret. Real system credentials belong in the Robomotion Vault, never in a flow.