Nov 21, 2026
/
Guide

Username Search for OSINT: How to Trace a Handle Across Platforms

How reverse username search works, why enumeration tools give false positives, and how to verify that two accounts with one handle are the same person.

Username Search for OSINT: How to Trace a Handle Across Platforms

Last verified: 21 November 2026.

When we built THINKPOL's free platform lookups, we first let them match a handle by name alone: if a Reddit post said "my Instagram is" followed by the handle, it counted. Then we had an automated check review 74 of those name-only matches. It rejected 29 of them. Thirty-nine per cent of plain name matches pointed at the wrong thing: a common word, a different account, a joke. We removed name matching from every lookup except six platforms that have no profile link to match on.

That number is the whole lesson of username search. A handle that appears in two places is a hypothesis, not a finding. This guide covers how reverse username search works, where the tools go wrong, and how to turn a match into something you can defend.

It takes one handle and asks where else it exists. Investigators use it because people reuse usernames: the name chosen for a gaming account in 2014 often turns up on a forum, a marketplace and a social profile years later. A reverse username search either checks hundreds of sites for an account with that exact name, or searches what has been written publicly about the name.

How do username search tools work?

Most open-source tools send a request to a profile URL on each site and interpret the answer. Sherlock, released under the MIT licence, describes itself as hunting accounts "across 400+ social networks". Its site documentation explains the detection methods: an HTTP status code, which it calls the most reliable, or an error message in the page, which it calls the least reliable. Maigret, also MIT, goes further: it extracts profile details and follows any new usernames and IDs it finds, recursively.

WhatsMyName took a different approach: a shared dataset of detection rules, 691 sites in its final version, under a Creative Commons BY-SA licence. Its rules require both an expected status code and an expected string in the page before calling an account found, and its README says to "never decide on the status code alone". The repository was archived and made read-only on 9 November 2026, so its detections will now decay as sites change.

Why do username searches give false positives?

Because "this URL returns a page" is not "this person has an account". Sherlock's own list of removed sites records the failure modes: sites that return the same page for claimed and unclaimed names, login walls that look like profiles, usernames with periods that break the check, and bot protection that blocks the request. On top of that sits the human problem. Short and common handles are taken on every platform by different people. A match on a seven-letter dictionary word means nothing.

Enumeration answers "does an account with this name exist", never "is it the same person". WhatsMyName's README puts it well: treat results "as leads to verify, not as facts".

How do you verify that two accounts belong to the same person?

Look for a link the person made themselves. The strongest evidence is an account that points to the other one: a profile URL in a bio, a post that says "my other account is", a shared link-in-bio page, a cross-post of the same image within minutes. After that come consistent details that are hard to fake together: the same rare interests, the same writing habits, the same time zone of activity, the same creation period. Each one alone is weak. Three independent ones together start to mean something.

Record the reverse direction too. If account A links to B but B never mentions A, someone may simply be borrowing a reputation.

Two reasons. First, Reddit usernames are stable: Reddit's help centre states that once a username is finalized, it can't be changed in any way, including capitalization. An archived comment from 2016 is still tied to the same account today. Second, Reddit users talk about their other accounts constantly: they share Instagram handles in selling threads, Discord invites in gaming communities, Telegram channels in crypto groups. Those are exactly the self-made links that verification needs.

Worked example: what a sample of handles looks like

While building the lookups, we sampled 314 handles and invite codes for 51 platforms from recent Reddit posts and measured each one across the archive. Of the 303 that returned results, 97 were mentioned only by name, never as an exact profile link. And 124 of the 314 were cited by a single Reddit account, against 135 that appeared in five communities or more.

Those two groups need opposite readings. A handle cited by one account, again and again, is usually self-promotion, which is a strong link between the Reddit account and the handle. A handle cited by many unrelated authors is usually a public figure or a brand, which tells you about the handle's reach but nothing about who runs it. The name-only third is where most of the false positives from our test lived.

How do you run a username search step by step?

Start with the exact string and its variants: with and without underscores and periods, and older forms of the name. Run an enumeration tool and keep only the hits you can open and read. Search the handle and its profile URLs in an archive of public posts, so you find what the person said about the account rather than only whether it exists. Then follow the strongest self-made link, and write down for each match why you believe it, with the date you checked. A deleted profile is not a dead end if an archived post recorded it.

What THINKPOL's lookups do not do

The free Reddit User Lookup returns a Reddit account's full history, and the platform lookups, such as the Discord Lookup or the Telegram Lookup, search THINKPOL's archive of 30 billion Reddit posts and comments for a handle from another platform, deleted posts included. They read Reddit only. They do not check whether an account exists on another site, never contact other platforms, and match exact profile links except on six link-less platforms. A result shows where a handle was written on Reddit, not who owns it. For Reddit-specific techniques, see Reddit username search.

FAQ

What is the best free username search tool?

For checking whether an account name exists, open-source enumerators such as Sherlock and Maigret are the usual starting point. For finding what was said about a handle, search an archive of public posts. Most investigations need both.

Can a username search identify a real person?

Not by itself. It produces candidate accounts. Identification needs links the person made between accounts and independent details that agree, recorded with dates.

Why does the same username show up on sites the person never used?

Common handles are claimed by different people on every platform, and some sites return a page for any name. Both create matches that look real and are not.

Sources

External, checked 21 November 2026:

  • Sherlock, README: 400+ social networks, MIT licence. https://github.com/sherlock-project/sherlock
  • Sherlock wiki, Adding Sites: detection methods and their reliability. https://github.com/sherlock-project/sherlock/wiki/Adding-Sites-To-Sherlock
  • Sherlock, removed sites and false positives. https://github.com/sherlock-project/sherlock/blob/master/docs/removed-sites.md
  • WhatsMyName, README: 691 sites, detection rule, archived 9 November 2026, CC BY-SA 4.0. https://github.com/WebBreacher/WhatsMyName
  • Maigret, README: recursive search, MIT licence. https://github.com/soxoj/maigret
  • Reddit Help, Can I change my username? https://support.reddithelp.com/hc/en-us/articles/204579479-Can-I-change-my-username

THINKPOL (21 November 2026): precision test of 74 name-only matches reviewed by an automated check, 29 rejected; sample of 314 identifiers for 51 platforms measured in the Reddit archive.

Questions or something to add? Discuss it in the THINKPOL Discord.

Share
THINKPOL

THINKPOL

Grey web intelligence

Investigations and analysis from the THINKPOL team, built on a 30-billion-post Reddit archive with deleted content preserved.

Newsletter

Grey web intelligence, in your inbox.

One email a month: new investigations, tool updates and what changed on the platforms we monitor.

Thanks for subscribing to our newsletter!
Oops! Something went wrong while submitting the form.
Only one email per month. No spam.