Most candidates do not get rejected. They get unread. The crisis of modern hiring is not the algorithm. It is the volume the human has to face.
Names are drifting through the field behind this essay. Move your cursor near one to slow it. Click to save. Saved names accumulate as a longlist further down the page — a living record of who you reached. The witness panel keeps your bandwidth honest.
The dormant data in any modern ATS is the largest unworked asset in talent acquisition. The applicant from six months ago, declined for one role, is sitting in your system right now perfect for the new one. Nobody has gone back to look.
This is not a metaphor. The figures are blunt. The average enterprise ATS holds tens of thousands of past applicants whose data quietly ages while new requisitions go to market and hire from cold. The cost of re-engaging a previously vetted candidate is a fraction of the cost of sourcing a new one. Almost no organisation does it systematically because nobody has the hours.
The first practical move available to most organisations is not to source more. It is to reread what they already have. AI is good at this. It is good at parsing a thousand stale CVs against a fresh requisition and surfacing the eight that newly fit. What it cannot do is the conversation that follows. That is still you.
— Names are drifting in the field above and below this card. Click any of them. The longlist further down is keeping the saves.
The economics of attention in recruitment are stark. There is enough time to glance at every application and almost no time to read any deeply. The question is where the deep reading goes.
A widely-repeated claim says 75% of resumes are rejected by Applicant Tracking Systems before a human ever reads them. The claim is wrong, or at least dangerously imprecise.
The myth has consequences. Candidates believe the door is automated, so they game keywords and despair when keyword optimisation does not work. The truth is harder: the door is human, and the human is overwhelmed. There is no algorithm to outwit. There is a person reading at the speed reading is possible, and the question is whether your application reaches them in the ten minutes a week they are not in another meeting.
Recruiters now manage 56% more open positions than they did three years ago, and they process 2.7× more applications. The bandwidth has not scaled. The mismatch is the story.
— Watch the names drifting past. The ones you click are saved. The ones you do not are counted as seen-but-passed in the panel below.
It is not in the CRM. It is not in the ATS. It lives in one person's head, built over years, and losing them is losing it.
The recruiter who has been on the desk for ten years knows things the system does not. Who is open to a conversation but not actively looking. Who has just had a baby and is six months from being approachable. Who would be brilliant in your team and not in the team next door. The system records what the candidate said about themselves on a Thursday at 2 p.m.; the recruiter records what the candidate is.
This is the part of the work that does not scale through automation alone, and it is also the part that compounds. Every conversation a senior recruiter has informs the next one. Every placement teaches them something about a category of fit. The mental map is a moat that takes a decade to build and a quarter to lose.
— The longlist below is the visible accumulation of your saves. It is the small, real version of what a senior recruiter holds in their head across years.
It is the panel doing its actual job — surfacing dimensions of fit that no individual can see alone. Calibration sessions do not eliminate disagreement; they make it productive.
There is a temptation, when AI hiring tools become more capable, to treat candidate evaluation as a problem of finding the right ranking algorithm. This treats hiring as if it had a single correct answer that disagreement obscures. It does not. A candidate is the meeting of a particular human with a particular role at a particular moment, and fit is irreducibly contextual. The same engineer can be the answer in one team and a poor match in another, and the difference is rarely in the engineer.
The data points one direction. Automation is welcomed where it removes friction from non-judgement work. Automation is rejected where it presumes to make a judgement only the human reading the room can make. The line between these is the recruitment professional's actual expertise, and it has rarely been more valuable.
— Two recruiters reading the same drift will save different names. That is the point. The diversity is the discipline.
Substance, function, capacity, effect. The expensive part of recruitment is reading the second pair well, and the second pair is what AI cannot do alone.
The CV. Roles, dates, employers, certifications. The countable history. Necessary and never sufficient.
The verbs underneath the titles. What they shipped, fixed, led, broke, learned. The texture of the work. Visible in the right interview, invisible on the document.
The latent shape — the role they have not held but would grow into, the second skill the first one is upstream of. Capacity is what good recruiters hire for and what bad processes systematically miss.
The mark on a team. The way the team would change shape around them. The legacy of the placement, evaluable only at twelve months and rarely measured. Effect is the candidate's true contribution, and it is a year out of phase with the hire.
— A process that reads only substance and function is a filter. A process that reads all four is a craft. AI handles the first pair well. The second pair stays human.
The longlist below is what your attention reached. The names that drifted past unsaved are the cost of the bandwidth. The reform recruitment needs is not better filtering. It is restored attention.
There is a way out of this, but it does not come from filtering harder. It comes from automating the friction and protecting the human moment. AI drafting first-pass acknowledgements within the hour. AI surfacing returning applicants whose substance and function newly match. AI clearing the administrative debt that absorbs the recruiter's day. The hours returned are the budget for the conversation, the calibration session, the four-axis read.
The Fiduci Recruitment position is the position the data already supports. Keep the human moment. Do not automate it away. Let the technology compose the friction and leave the meeting in the room. The candidates who arrive in your column when the door is human are the candidates who stay.
— Below is your longlist. It is alive. The proportion you reached is the work.
A living longlist. Each row is a candidate you clicked from the drift. Anonymised, fictional. The proportion you reached is real.
The names above are not real. The proportion you reached is. If your bandwidth here is below 10%, you are reading at the rate of a working recruiter. The argument of this essay is the proportion.