# Rapport Contact Deduplicator

> Find duplicate contacts in a CSV (same e-mail, phone or similar name) and merge them. Checks a contacts CSV (Rapport, HubSpot, Pipedrive or any file with name, e-mail, phone and company columns) for duplicates: the same e-mail even with different case, Gmail dots or +tags, the same phone number written differently, and very similar names at the same company. Shows every group with its reasons and gives you a merged CSV. Same rules as Rapport’s duplicate finder.

Part of usefulweb.app (https://usefulweb.app/). Part of [Rapport](https://rapport.usefulweb.app/) (ownerApp: rapport). Results are deterministic calculations from the inputs and the listed assumptions; the API returns exactly what the web page shows.

## Use it
- Human interface: https://rapportcsvdedupe.usefulweb.app/ (interactive tool at https://rapportcsvdedupe.usefulweb.app/tool)
- JSON API: POST https://rapportcsvdedupe.usefulweb.app/api/v1/calculate with a JSON object of inputs (schema: https://rapportcsvdedupe.usefulweb.app/api/v1/tool)
- OpenAPI: https://rapportcsvdedupe.usefulweb.app/openapi.json
- MCP (Streamable HTTP): https://rapportcsvdedupe.usefulweb.app/mcp - tool `rapport_csv_dedupe`

## Inputs
- csv (textarea): Contacts CSV - With a header row. Names, e-mail, phone and company are used; other columns are kept in the merged file when Rapport knows them.
- csvFile (file): CSV file (optional) - Used instead of the pasted text when given.

## Method
- Same e-mail: compared in lower case; for every provider a +tag is ignored (anna+news@ = anna@), for Gmail also dots (an.na@gmail.com = anna@gmail.com).
- Same phone: only digits count and the last nine are compared, so +49 40 2345678 and 040 2345678 match. Numbers with fewer than seven digits are ignored.
- Similar name: Jaro–Winkler similarity of the folded full names (accents, case and punctuation removed) of at least 0.92, AND the same company (legal forms like GmbH, Inc. or B.V. ignored) or the same non-free-mail e-mail domain. Identical names without any company or e-mail on one side are flagged as weaker matches.
- Pairs are joined into groups. In the merged CSV the first row of each group is kept; its empty fields are filled from the others, tags are combined, notes appended and the newest documented lawful basis/consent wins.

## Limits
- Check each group before you delete anything in your CRM: two real people can share a phone number or a name.