The challenge
The company keeps its environmental records to a strict corporate naming standard, set out in a multi-page written document. It is scanning decades of historical paper.
The approach
Statvis reads each document itself, not just its old file name, and applies the rules in the company's written standard to name it. The same rules are applied the same way to the first document and the 78,000th. People get tired and drift; an automated workflow does not.
Names that the standard can't settle from the document are flagged for a person to decide, rather than guessed.
How we tested it
We ran a blind test on 251 of the company's documents. Each document was named by Statvis and by people, and every name was scored against the company's written naming standard without the scorer knowing which was which. Statvis made 13% fewer naming errors than the people did.