Le texte intégral de ce guide n'est actuellement disponible qu'en anglais.
Start with the handle, not the name
A username is the most durable identifier most people leave behind. Names get shortened on one platform and spelled out on another; emails get abandoned; a handle someone picked in their teens tends to follow them for a decade because it is the thing their friends recognise.
So before you search anything, write down the exact string. Underscores, dots, trailing digits and capitalisation all matter. Most platforms treat handles as case-insensitive but punctuation-sensitive, which means "jenny_oslo" and "jenny.oslo" are two different people until proven otherwise.
Run the scan, then read it properly
A cross-platform scan takes that one string and checks it against the public profile pages of hundreds of sites at once. What comes back is a list of pages that returned a profile for that exact handle — not a list of pages that belong to the person you have in mind.
That distinction is the whole game. A hit means "this handle exists here". It does not mean "this is the same human". Common handles belong to different people on different sites, and the more ordinary the string, the more false links you should expect.
Verify the hits yourself
Open each match directly rather than trusting the row. Check whether the avatar, bio and posting history are consistent with what you already know, and whether the account looks established or freshly made. A handle registered last month with three posts is a different signal from one with years of history.
If two profiles are supposed to be the same person, look for a link one of them made public: a website field, a cross-posted photo, a shared alias in the bio. Evidence that the person published themselves is worth more than any similarity you infer.
When the handle comes up empty
Absence is information too. If a handle resolves nowhere, either the person uses a different string per platform, or the accounts are private enough that public checks cannot see them. Try plausible variants — a first name plus a year, the same handle with an underscore — and stop when the guesses stop being evidence.
Never let an empty result become a conclusion about the person. It only ever describes what a public scan could see, not what exists.