Username Sweep vs Deep Search

These look like the same feature with different names. They are not. One asks the platforms; the other asks the search engines. They fail in opposite directions, which is exactly why both exist, and why running only one of them leaves a predictable kind of gap.

How the sweep works

The sweep holds a catalogue of platforms and a URL pattern for each. It builds the profile address for your handle, requests it, and reads the answer, preferring a platform's own public API where one exists, because an API that returns a user object or a documented 404 is an unambiguous yes or no in a way that a web page never is.

The hard part is not making the requests. It is deciding what an answer means. A great many sites return HTTP 200, “here is your page” for a handle nobody owns: a generic template, a “no results” screen, a sign-in wall. A checker that treats a 200 as an account will hand you a confident list of profiles that were never there. So each platform lands on one of three verdicts, confirmed, not found, or could-not-tell, and the third is reported rather than rounded into one of the other two. The measurement behind that design is in why username checkers report fake accounts.

How deep search works

Deep search has no catalogue. It queries search engines with carefully built operator strings, the same technique as advanced search operators, applied systematically, and reads what comes back. Because it is not restricted to a list of known platforms, it can surface a forum post, a pasted document, an archived page or a small site nobody would ever put in a catalogue.

Its weakness is the mirror image of that strength: it only knows what has been indexed. A search engine has to have crawled the page and decided to keep it. A brand-new profile, a site that blocks crawlers in robots.txt, anything behind a login, and anything simply never linked to from elsewhere is all invisible to it, while being perfectly findable by asking the platform directly.

Side by side

What a sweep result actually looks like

Each platform comes back as one of three things, and the difference between them is the point of the whole tool.

• Confirmed. Something specific was established: the platform's own API returned a user object for that exact handle, or the page carries profile markup naming it, or the response demonstrably differs from what that same platform returns for a handle nobody owns. Each confirmed row says which of those it was.
• Not found. The platform gave a real negative, a documented 404, or its own “no such user” page. This is an answer, so it is acted on rather than shown: there is no value in a list of four hundred places you do not have an account.
• Couldn't verify. The platform answered, but nothing in the answer settles it. It served a bot check, or a sign-in wall, or it is one of the sites that shows the same page whoever you ask for. These are listed with the reason attached and are never counted as accounts.

That third category is where most tools quietly cheat. Rounding “couldn't verify” up to “found” produces a longer, more impressive list; rounding it down to “not found” produces a reassuring one. Both are claims the evidence does not support, and the result summary says how many platforms landed there so a thin result cannot be mistaken for a clean one.

What a deep search result looks like

A ranked list of places the handle appears, each with the context it appeared in. There are no verdicts, because there is nothing to have a verdict about, a search engine returning a page is not a claim that the page is about your person.

Judging that is your job, and the useful questions are always the same three: does the surrounding context match what you already know, does the handle appear as an identity or merely as a string, and does the same detail show up in more than one independent place. A handle that appears once, in a comment, proves very little. The same handle on a profile that links to another profile you have already confirmed is considerably stronger.

A worked example

Say you are cleaning up an old handle you used at sixteen and want it gone.

The sweep is the right first move, and it is nearly the whole job. It returns the platforms where that handle currently resolves to an account, each as a clickable address. That list is directly actionable: open each one, log in if you still can, delete or lock it. Anything in the “couldn't verify” section is worth one manual glance, because a sign-in wall is often exactly what an account you still own looks like from the outside.

Deep search answers the different question you will have afterwards: what did that handle leave behind that deleting the account will not remove. Forum posts quoted elsewhere, a profile captured by an archive, a document with the handle in its metadata. Deleting an account does not un-index a page that copied it, and that is the gap the sweep structurally cannot see. Our guide on reducing your digital footprint covers what to do with what you find.

Three mistakes worth avoiding

• Reading an empty deep search as “nothing exists”. It means nothing was indexed and kept. Those are very different, and the mistake runs in the comforting direction.
• Assuming one handle is one person. Common handles are shared by unrelated people constantly. A set of accounts with the same name is a set of candidates, not an identity, and treating it as one is how people get misidentified.
• Trusting a long list over a labelled one. A tool returning ninety hits where this one returns thirty has not found more; it has usually counted every page that loaded.

Which to run

Start with the sweep. It is faster, it gives a definite per-platform answer, and for the most common job, “where does my old handle still have accounts”, it is the whole answer. The output is a list you can act on: each row is a real address you can open, close or clean up.

Then use deep search for context. It is the better tool when the sweep has already told you a handle is in use and you want to know what is attached to it, or when the handle is distinctive enough that indexed mentions are likely to be the same person. It is the weaker tool for a common handle, because the results will be a mixture of people who happen to share a string.

The trap worth knowing. Deep search tells you what search engines have indexed, not what exists. An empty deep search is not evidence that a handle is unused; it is evidence that nothing about it was crawled and kept. Reading one as the other is the most common mistake people make with this kind of tool, and it runs in the reassuring direction.

Why they are separate tools

It would be possible to merge them into one button and present a single list. We have not, because the two kinds of result mean different things and mixing them makes both less useful. “An account exists at this address” and “this string appears on a page somewhere” are not the same claim, and a combined list would quietly flatten the difference, which is the failure mode this whole tool is built to avoid.

The same reasoning shapes how results are labelled throughout: a platform that was rate-limited is shown as unchecked rather than clear, and a breach lookup that could not reach its sources says not checked rather than no breaches found. More on that in the six tests.