The checks it runs

Periscope runs three checks on every document, in order. Each one only runs if the check before it passed. If an early check fails, the result tells you where it stopped, and the checks after it simply do not run.

1. Is it a document?

The first check is basic on purpose: is this a readable document at all, not a blank page, a random photo, or an unreadable scan. Nothing else can happen until this passes. If it does not, the result stops here and says so.

2. Does it match the type you chose?

If the image is a document, Periscope compares what it reads against the expected shape of the document type you selected. A passport looks different from a bank statement, and a bank statement looks different from a utility bill. This comparison produces the Document match score: how closely the document fits the type you told Periscope to expect.

3. What can we read?

If the type matches, Periscope extracts the fields that document type should have: names, dates, numbers, addresses, whatever applies. What it finds, and how clearly, becomes the Evidence behind the result. A field that is missing or unreadable shows up as exactly that, not as a silent gap.

Why the order matters

Running the checks in this order means a bad upload fails fast and clearly. A blank page never gets scored against a document type. A document of the wrong type never gets its fields extracted and reported as if they meant something. Each score you see reflects a check that actually ran, on a document that earned its way there.

For how to read the scores themselves, see Reading your result. Periscope does not check whether a document is genuine or forged; that is covered separately in What Periscope doesn't check yet.