Extract the data from an insurance declarations page, coverage by coverage

What comes back

You email a US declarations page, as a PDF or a phone photo, and get a sheet back as a PDF, usually within a few minutes:

• Every coverage as a row, with its limit, deductible and premium side by side.
• Deductibles printed for the policy as a whole in their own block, such as a hurricane deductible of 2% of Coverage A, instead of being squeezed into one coverage row.
• The key fields: insurer, named insured, policy number, policy term, property location, total premium.
• A mark on every value: ok, review, flag or missing. What needs a look is listed at the top, with the reason and the page it was read from — and the line, where the file has text.

Why not just run it through OCR?

OCR gives you the text of the page. You still have to find the coverages in it, match each limit to the right line, and notice when a number was misread. Every carrier lays its dec page out differently, so a template that works for one breaks on the next.

DecTrace reads the page itself, not a carrier template. Each page is read three times independently. Where the readings disagree, or a value cannot be tied to a line on the page, the value is marked for review instead of passed along as if it were certain. A page it cannot read reliably comes back saying so, rather than as a confident-looking table.

Where agents use it

• Quoting a new client who brings their current policy: the coverages are in front of you without re-keying them.
• Remarketing a renewal across carriers: the current program as a table to compare against.
• Checking what a client actually has before a coverage conversation, from the photo they texted you.