Nobody good is browsing your job post

Two hundred applicants, none of them right, and the person you actually want is employed.

This is the second search that has not worked. You rewrote the description. You added competitive equity. You paid to promote the post. The applicant count went up and the shortlist did not move, and screening the difference cost your team a fortnight it did not have.

Meanwhile the three people you would hire tomorrow are all doing good work somewhere else. None of them has read your post, because none of them is reading posts.

The old line about this is that nothing can reach those people, and that line is now out of date. Indeed Smart Sourcing advertises more than 370 million sourceable profiles, and Talent Scout, launched in September 2025, drafts outreach to people who never applied. Passive candidates are reachable, and pretending otherwise would be easy to check and wrong.

What still holds is what all of it is working from. A post reaches people who are looking. Sourcing reaches people who filled in a profile and set a flag, which is a description of somebody’s past plus a declared availability. Neither of them knows that your head of engineering has been in a thread with the right person for two years. And that, far more than the wording of the message, is what decides whether it gets read or archived.

You do not have a funnel problem. A bigger funnel is the problem.

Every widening of the funnel adds people who are not right, and makes the person who is right harder to see. It also quietly changes what your team spends its week doing, from judging a few candidates carefully to eliminating many candidates quickly, which is a different job requiring a different temperament.

The alternative is to start from a want rather than an advert. In Deals On, function and seniority are separate dimensions on purpose. Fold them into one string and “senior backend” stops matching “backend, senior”, which is the kind of failure nobody sees, because a search that misses does not report what it missed.

Here, unlike co-founding, overlap is rewarded rather than punished. It is the same engine, deliberately inverted: you want a co-founder who is not you, and a backend engineer who is.

The other half of it belongs to the person being hired. Saying what you want next is private. It surfaces only against somebody actually hiring for it, which means a person can declare a direction without announcing it to their current employer, their team, or anyone browsing.

One search, from both sides.

On day zero she does not write a job ad. She writes a want: senior backend, remote or Berlin, this band, starting in Q1.

That night it is matched against people who have privately said what they want next. None of them has said so in public, which is usually true of the people worth hiring.

In the morning there are four people. Two of them are reachable through somebody she already knows.

Two weeks later the introduction goes through the person who can vouch, rather than as a cold message from a stranger. She interviews three people instead of screening two hundred.

She never posted the role. The funnel was the problem.

This does not make hiring easy, and anyone selling you that is selling something else. Interviews still have to be good. References still have to be taken. What changes is the quality of the four conversations you start with, and how much of your team’s judgment is still intact by the time it gets to them.