"This is LinkedIn Recruiter," I told him, and turned my screen round so he could watch. "This is how recruiters like me look for people like you, every day."
The search starts with a job title. The ads for the work he had been doing for ten years kept using the same one: Operations Manager.
His profile did not contain those two words anywhere.
So I told him the thing I end up telling most people in that chair. Recruiters type a job title first, always. It is the single most reliable way we find anyone, and if the title on your profile is not the one we type, we are not going to find you.
That part is easy to fix. What fixing the title does not fix is the rest of the problem, and I did not understand the rest of it until I put the question to a machine.
Filling one role does not involve one search. It involves several structurally different ones, each starting from a different kind of evidence: the job title, the specialist skill, the current or previous employer, and the shape of the team someone runs. Each is built to reach people the others would miss. Recruiters have always had these routes available; the agentic sourcing tool I tested runs all of them at once, while a busy human usually runs only the title search. Either way, a profile built around a job title can be found by that one and is invisible to the rest, so whether you are seen depends on which search happens to get run.
Earlier this month I gave a real Sydney engineering brief to one of the agentic AI sourcing tools that now do this work end to end, and five minutes later it handed back a ranked shortlist built from public profiles. I wrote that session up here.
A few days later I went back with a list of questions. One of them was a question I had never thought to ask a person: how did it decide who to look at in the first place? I have not found the answer written down anywhere.
"Good angles search the same brief through different evidence; redundant angles search the same evidence with different wording."
An angle, in its language, is a route into the same shortlist. The requirements never move. What moves is the evidence it uses to go looking. And it was blunt about synonyms: rewording a search you have already run just runs it twice.
For that one Sydney role, it ran four.
| Search | Where it starts | Who it reaches that the others miss |
|---|---|---|
| Title | Engineering Manager, Senior EM, Software EM, Development Manager | Anyone whose title is recognisable, even if their profile says nothing else. Broadest reach, shallowest evidence. |
| Specialism | Design systems, component library, design tokens, UI platform, accessibility infrastructure | The specialist whose formal title is just "Engineering Manager" and whose real expertise lives in the role description. |
| Employer | Engineering leaders currently or previously at the large product companies where this work concentrates | The recently promoted person whose current title is unusual, but whose employment history is the giveaway. |
| Scope | Managing managers, leading several teams, coaching engineering managers, group responsibility | The leader who writes about organisational work and never once uses the specialist vocabulary. |
Read the right-hand column again, because that is the whole mechanism. Each of those four searches exists to catch someone the other three would drop.
Here is the part that should worry you more than it comforts you. A tool runs all four because running searches costs it nothing. A human recruiter with six roles open and a shortlist due Thursday usually runs one, sometimes two, and it is almost always the title. So the machine version of this is the generous one. Whether you get seen depends on which search gets run, and the fewer of them you can be reached by, the more of your career is decided by somebody else's Tuesday afternoon.
The tool gave me a test for whether a search is genuinely separate or just a reworded version of one it already ran: could a qualified person turn up through this route while being plausibly absent from every other? If the answer is no, you have reworded a search rather than added one.
Here is the thing worth naming, because once you can count it you can fix it. Call it your search coverage: how many of those four searches can actually reach you.
Most senior profiles I open have coverage of one. The title.
That is not laziness. Every piece of advice they have ever been given points there. Rewrite your headline, add keywords, use the right job title, and every one of those instructions improves the same single search. You end up with a thicker version of the route you already had, and three routes that still run straight past you, and no way of knowing that is what happened because the outcome of a search you were never in looks exactly like the outcome of a search you lost.
The other three searches read the parts of your profile nobody re-reads. Your specialism is proven by what you wrote under your current role. A skills tag only asserts it. Your employer history is a route you already own and never think about. Your scope, the size of the team, whether a manager reports to you, what you actually own, is the rarest evidence of all, because almost nobody writes it down.
Which makes it the most valuable thing you can add. Near the end of that second session the tool volunteered a principle I have not stopped thinking about: "A candidate's position depends partly on who else satisfies the brief." Ranking is relative. Which means the evidence worth adding is whatever everyone else in your pool leaves out, however unglamorous it looks to you. If you are the only person in the pool who states their team size and reporting line, that one sentence does work no amount of polish on your headline could buy.
Open your profile and answer four questions honestly. Each one is a search.
1. Would a title search find you? Look at three live ads for the job you want and note the exact title they use. If your profile does not carry that title in your headline or a recent role, you fail this search. If your title is internal, keep it and add the plain equivalent next to it: "Chapter Lead, Platform (Engineering Manager equivalent)". That clarifies a job you actually hold.
2. Would a specialism search find you? Find the two or three specialist terms that appear in those same ads. Now check whether they appear anywhere in your experience section, attached to a role and a date. A skills tag does not count. If the term only exists in your skills list, you show up as a term with no evidence underneath it.
3. Would an employer search find you? Two parts, and most people only do the first. Check every company in your history is named the way that company writes its own name, which takes ninety seconds. Then check whether any of them is a place a recruiter for your target role would actually think to search. Accurate names are the precondition. The route only exists for you if your history contains companies that are known for the work you are chasing, which is why this search rescues some people entirely and does nothing at all for others.
4. Would a scope search find you? This is the one almost everyone fails. Write one sentence under your current role that says what you run: how many people, whether any of them manage others, what you own, and how far it reaches. "Manage eight engineers across frontend and backend, including one team lead, responsible for the payments platform used by four product teams."
If the work is confidential, describe the shape without the specifics. "Lead a nine-person team across frontend and backend, including one manager, delivering shared infrastructure across several enterprise products" names nothing sensitive and still answers every question a scope search asks.
That fourth sentence is worth more than ten skills and a week of posting. It is also the change I most often find missing when I open a senior profile, and writing it down makes nobody more qualified than they already were. They stopped being reachable one way only.
Posting and activity are a separate mechanism, worth doing for different reasons. If you want the detail on why they do nothing for this particular pipeline, I wrote that up here.
I want to be straight about the limit of this, because the advice industry never is.
Coverage is one of three problems, and it is only the first. If your profile is missing titles, terminology or location, that is a visibility problem and everything above fixes it. If your profile is findable but thin, no scope, no ownership, no outcomes, that is an evidence problem and the fix is writing down work you have already done. But if the role you are targeting genuinely requires experience you have not had yet, no rewrite touches that, and making yourself more findable will only make the mismatch more visible to more people.
That third one is a conversation I have had more than once with people who came to me asking for a better headline. Nobody wants that answer. It is usually the one that saves them months of aiming at the wrong thing.
Careersy AI's AI Discoverability check runs the outside read for you: it scores how visible you are to the AI sourcing tools recruiters actually run, names what is keeping you out of the results, and shows you what to change to start showing up. I built it because I kept doing this read by hand in sessions and finding the gap in the same place every time.
In a later session he asked whether we were even allowed to change how his roles were described. I told him what we were actually doing. We were translating the internal titles to the external market, so the external market could understand them.
That is the entire job. Every word of it was already true about him.
He also asked how long his clients take to do this. About an hour, I said, because everything the searches need is already sitting in your CV.
Four searches, and you are probably reachable by one. Working out which one takes ten minutes. Fixing it takes the hour he was quoted, and the three that have been running past you stop.
Check your profile against the searches recruiters run →Filling one role takes several structurally different searches rather than one. Four common ones start from the job title, a specialist skill or technology, the candidate's current or previous employer, and the scope of the team the candidate runs. Each is designed to reach qualified people the others would miss. The agentic sourcing tool tested for this article ran all four for a single role; a recruiter under time pressure usually runs only the title search. That is why the same person can be highly visible in one search and completely absent from another.
Usually because your profile only supports one of the searches being run, most often the title search, and your job title does not match the words recruiters actually type. Internal or invented titles are the most common cause. Check three live ads for the role you want, note the exact title they use, and make sure that title appears in your headline or a recent role.
It helps you appear. It does not help you qualify. A skill tag can surface your profile in a search, but it carries no evidence of when you did the work, at what level, or what it produced. The same term written into an experience bullet, attached to a role and a date, is what moves you up the list once you have been found.
One sentence under your current role stating what you run: team size, whether any of your reports manage people, what you own, and how many teams or products it reaches. Very few profiles state reporting lines or team size, so it is both the rarest evidence and the hardest requirement for a recruiter to verify any other way.
No, and you should not misrepresent it. Keep the official title and add the recognised equivalent alongside it, for example "Chapter Lead, Platform (Engineering Manager equivalent)". Then explain the team and reporting structure in the role description. That makes an accurate title searchable without claiming a job you do not hold.
Not in the searches described here, which read your profile rather than your activity. Posting reaches people directly, which is a genuinely useful but completely separate mechanism. The longer answer, including what Open to Work and profile-update timestamps actually do, is in this piece on AI discoverability.
An agentic AI sourcing tool told me exactly how it ranks candidates. Three separate jobs, not one, and posting more does none of them.
1 min read
We ran the exact questions people bring to intake calls through ChatGPT, Claude, Gemini and Grok, then through Careersy AI's new AI Fluency mode on the same rubric. Zero of twenty answers asked what was actually going on before advising, and three of the four told users the AI pay premium is Australian. It's global.
11 min read
