Coverage and accuracy

Coverage by input type

Platforms checked per lookup

887

Reachable from an email address

27

Reachable from a username

864

Reachable from a phone number (beta)

1

Instant tier, email lookups

25

Instant tier, username lookups

25

Platforms with their own page

117

Most platforms are keyed on a username, which is why a username reaches the widest set. Email reaches fewer platforms but the highest-signal ones - the accounts people actually register with their real address. Phone coverage is in beta: registration by number is exposed by very few platforms, so a phone lookup leans on number validity, carrier and line type rather than a long list of accounts.

Coverage by category

• Social media13 with a page
• Developer25 with a page
• Creative12 with a page
• Media & publishing23 with a page
• Gaming5 with a page
• Shopping & marketplaces9 with a page
• Payments & crowdfunding4 with a page
• Professional & productivity8 with a page
• Communities & forums11 with a page
• Accounts & identity7 with a page

Counts are of platforms with their own page - the recognisable ones. The long tail of forums, regional sites and niche services is checked in the same request; the full list is on the platforms page.

Beyond registration: breaches and web mentions

Every lookup also returns breach history for an email address (which breaches, when, what classes of data) and public web mentions (pages where the identifier appears). Both come from dedicated sources, are cached separately, and are null rather than an empty list when the source is unavailable - so “no breaches found” and “could not check” are never confused.

How accuracy is verified

Two failure modes are tested separately, because they are different problems. Fabrication: does a platform ever report an account for an identifier that has none? Every platform is run against inputs known to have no account there; any positive is a fabricated result and fails the run. Discrimination: does a platform ever miss an account that is known to exist? Every platform is run against inputs known to have an account there; any miss is a false negative and fails the run. On 2026-08-29 the harness covered 887 platforms with 0 fabricated positives and 0 false negatives.

A platform that cannot pass both tests is not shipped. A platform that later starts failing them is removed from the registry until it passes again - which is why the response omits a platform it could not confirm instead of guessing. The date above is the date of the last full run; it is updated when the run is.