Most indexing mistakes are small: a transposed digit, a missing letter in a surname. They are also exactly the kind of mistake a second reader misses, because people tend to see what they expect to see.
Verification in RecordOne is designed around that.
The verifier types, not just reads
A county chooses which index fields are blind. In the Verification queue, those fields start empty. The verifier reads the value from the page image and types it, without seeing what the indexer entered.
The comparison happens on the server
RecordOne compares the two entries. If they match, the field is confirmed. If they do not, the field is flagged, and the verifier is told only that it does not match. The indexed value is not revealed, so the verifier goes back to the document, not to the other person's answer.
The county decides how much
Not every field needs this treatment, and not every document type needs a second person. The county sets which fields are blind and which workflows include verification. Digitized historical documents, for example, skip verification unless the county turns it on.
After verification
Once a document is verified it is published to Document Search. If something is still wrong later, a reindex task corrects the published information without changing the document's recorded identity.