← All articles

Under the hood

Reading an Oracul report: passport, graph, timeline and account decisions

The scan is done, now what? A walk through the report section by section: what completeness means, where footprint age comes from, why rejected accounts are kept and how to fix the profile when the engine is wrong.

An Oracul report puts the answer on top and the evidence below. The first screen says who the person is and how fully we see them. Below is what that rests on: every account, every link and every decision with its reason.

Passport and completeness

The passport collects what accepted accounts confirm: names, usernames, emails, phones, location, timezone, work, links and avatars. A username makes it into the passport only if an accepted account stands behind it.

Completeness is the share of ten fields: nine passport fields plus at least one high-confidence account. Empty fields come with a hint on how to extend the scan, such as adding a phone or a full name.

Summary

A short prose portrait: what the person does, their interests, where and when they are active. An AI model writes it from already filtered data. The "Target / round 1 / round 2" switch shows what the report would look like had the scan stopped earlier.

Link graph

The graph shows who was found through whom and in which round: the seed username, an email from its commits, a Telegram handle from the bio. It shows how far each find lies from the seed data, and where a chain might have turned to another person.

Timeline

Registration and activity dates of all accepted accounts on one scale. The earliest date is the age of the digital footprint. In antifraud it is one of the key numbers: a real person usually leaves a trail over years.

Accounts: accepted and rejected

Every account has a confidence level, a decision and evidence: what matched and what went into the profile. Rejected accounts do not vanish; they sit in a separate list with a reason.

  • namesake: a person with the same name.
  • name_mismatch: someone else's name on the account.
  • shared_nick: several different people use this username.
  • face_mismatch: a different face on the avatar.
  • excluded: the analyst marked it "not them".

When the engine is wrong

Mark an account "this is them" or "not them" and the profile is recalculated. "This is them" makes the account a reference; "not them" removes it and everything found only through it. The raw scan result is never changed, so any mark can be undone.

Read next

Run your first scan today

5 free scans every month. No card required.