Match
Resolve your list against another and find every true pair, even with messy names, typos, and formatting drift. Then merge each match into one clean golden record.
See it live
Two messy lists in, one clean file out.
Dirty lists cost reps real selling time.
Sales ops and back-office teams constantly pull and compare lists: a CRM export against a master account file, a new lead batch against existing customers, two regional books that overlap. The data never lines up. Names are spelled three ways, addresses are half-filled, the same company shows up as five records.
Generic dedup tools either choke on the volume or flag everything as a maybe. ProxiMatch is built for this job. It resolves entities with real probabilistic scoring.
What it does, in order.
Here's the whole cleanup in five moves.
- 1
Drop in your messy list.
Upload the file you actually have: the CRM dump, the distributor list, the spreadsheet three people edited.
- 2
Match it against your reference.
ProxiMatch resolves your list against another and finds every true pair, even when the names do not look alike. Six algorithms and term-frequency scoring separate a real match from a lucky coincidence. Every pair gets a confidence tier.
- 3
Verify and merge.
Run Verify to clean a list on its own: dedupe, standardize addresses, flag the stale rows, and merge each duplicate cluster into one surviving record. It runs by itself or automatically inside a match.
- 4
Fill in what's missing.
Forge appends what your list is missing: geocode, NAICS, specialty and more.
- 5
Export a file you can trust.
Take the clean, CRM-ready file out with a match-weight breakdown behind every pair. You can show exactly why two records were called the same.
One engine, three jobs, one clean file. Priced flat. No per-record fees.
What's in it.
One resolution engine, three jobs.
Verify
Clean a single list: dedupe, standardize addresses, and flag stale rows. Runs on its own or automatically inside Match. It can also merge each duplicate cluster into one surviving golden record.
Forge
Enrich your list with the fields it's missing, like geocode, NAICS and specialty.
Five confidence tiers
Every pair scored, sorted, and explained.
A six-algorithm stack scores each candidate and drops it into one of five tiers, from auto-accepted to needs review. You decide the edge cases. ProxiMatch handles the obvious ones.
Six algorithms
Real scoring, not one fuzzy slider.
Damerau-Levenshtein, Jaro-Winkler, N-Gram cosine, Soundex, Soft-TFIDF, and UID matching combine into one confidence score per pair.
- MinHash LSH blocking to compare large lists without the Cartesian blowup
- Fellegi-Sunter probabilistic scoring
- Five confidence tiers: auto-accepted, high confidence, needs review, conflict, unmatched
- Account type awareness: a differing or misspelled account type flags a pair for review instead of dropping a real match
- Accept and reject in-browser, with decisions tracked in the export
- CSV export mapping resolved matches back to your original IDs
ProxiCenter comes with every line.
BuildReports from any list you saved.
DesignDashboards you lay out yourself.
ShareYour data out to your CRM or your own tools through the API.
Your lists stay yours.
Everyday files run in your browser and never leave your machine. Big jobs run on our server, and your uploaded lists are deleted within two days of the match finishing.
$49 a month.
Match your lists and see every result. $49 a month to export, save, and run big jobs on our server.
FAQ
Common questions.
How big a list can it handle?
It uses MinHash LSH blocking to avoid comparing every record against every other. In your browser, a run handles up to 25,000 rows a list and 4 million comparisons, about 2,000 rows against 2,000. Anything bigger runs on our server.
What do I get out of it?
A reviewed, CRM-ready CSV that maps each resolved match back to your original record IDs, with a confidence tier on every pair.