Match inconsistent data. Enrich every record. Cut costs. Drive revenue.

What to do with a similarity key

One call returns one key. The matching happens when you compare keys across your records.

1. Generate a key for every record. Call a matching API for each value, or let a wizard or batch job process a whole file or table.
2. Group or join on the key. Records with the same key form a candidate group. Join two tables on the key to match across systems where the text never lines up exactly.
3. Review the candidate groups. A shared key flags a likely match, not a guaranteed one. Check each group against the other fields you hold, such as address, domain, or IDs, before you update or combine anything.

Working example: match a table from a notebook How similarity keys support entity resolution

-- Candidate groups within one table
SELECT simkey, COUNT(*) AS records
FROM customers
GROUP BY simkey
HAVING COUNT(*) > 1;

-- Likely matches across two systems
SELECT c.company, v.vendor_name
FROM crm_accounts c
JOIN erp_vendors v
  ON c.simkey = v.simkey;

Once each row has a similarity key stored beside it, plain SQL does the rest.