The Role Has Been Open 90 Days: How to Find the Leak and Fill It

A role that has been open for 90 days is almost never short of candidates. It is stuck at one stage, and you can find which stage in about an hour by pulling four numbers: how many qualified people entered the pipeline, how many replied, how many reached a first interview, and how many were rejected after that first interview. Whichever number collapses is the actual problem. Fix that one, reset the definition of the hire with the hiring manager, and you stop buying another quarter of the same result.

The reflex when a req goes stale is to add volume. More names at the top, another sourcing tool, another agency on contingency. That is the most expensive available response, because if the leak is at reply rate or at hiring manager calibration, everything you pour in the top runs out the same hole. You pay for the sourcing, and then you pay again for the weeks it did not buy you.

Run the funnel autopsy before you add anything

Pull these five numbers for the entire life of the req, not the last two weeks:

1. Qualified people contacted. Not names on a list, people someone actually judged plausible for this role.
2. Replies of any kind, including "no thanks."
3. Screens completed.
4. First interviews with the hiring manager.
5. Rejections after that first interview, with the stated reason written down.

Do not compare these to an industry benchmark. Compare them to your own last three filled roles at a similar level. Your company converts the way it converts, and your own history is the only honest baseline you have. The stage where this req looks nothing like the three that closed is the stage that is broken.

One caveat that saves a lot of wasted effort: if you cannot reconstruct these numbers at all, that is the finding. A role you cannot audit is a role you are managing on vibes, and 90 days is what managing on vibes costs.

Leak 1: not enough qualified people ever entered

The symptom is a small contacted number and a reasonable reply rate. The people you reached were interested. There just were not many of them.

This is the only leak where more sourcing is the right answer, and even here the fix is usually definition before volume. Widen one dimension at a time and watch what it does to the qualified pool: adjacent titles, adjacent industries, a wider radius or full remote, minus two years on the experience band. Change one, count the pool, then decide. Widening four things at once gives you a bigger list and no idea which change made it bigger.

If the pool is genuinely thin at every setting, you are hiring into a market where the people you want are employed and not looking, which is a different job than posting and waiting. That is the case for going after the candidates who never apply directly rather than hoping the applicant flow improves on its own.

Leak 2: they see the message and do not reply

Plenty contacted, almost nobody responding. This is the most common 90 day failure and the one most often misdiagnosed as a market problem.

Single digit reply rates on passive outreach usually mean one of three things. The targeting is wrong and you are messaging people the role is a lateral move for, or a step down. The message leads with your company instead of with the specific reason you contacted this specific person. Or the role has no visible answer to "why would I leave a job I am fine with," which is the only question a passive candidate is actually asking.

The practical test: read your last outreach message and delete every sentence that would still be true if you swapped in a competitor's role. If more than half the message survives, it is not personalized, it is templated with a first name at the top.

A line that works better than most openers, because it is specific and it names the tradeoff honestly:

"You have spent the last three years on [specific thing from their work]. This role is the same problem with a bigger budget and no on call rotation. Worth fifteen minutes to see if the tradeoffs land?"

Leak 3: they reply, then go quiet before the first interview

Good reply numbers, thin interview numbers. This is process drag, and it is entirely yours to fix.

Count the days between a candidate saying yes and a calendar invite existing. Then count the days between the screen and the hiring manager conversation. A passive candidate who agreed to talk was interested for roughly the length of the conversation that made them interested. Every day after that, their current job pulls them back. Slow scheduling does not lose you the candidate at the moment of the delay. It loses you the candidate three weeks later when they stop replying and you never learn why.

The fixes are boring and they work: hiring manager slots held open weekly rather than requested ad hoc, a scheduling link that goes out in the same message as the yes, and a rule that no candidate waits more than 48 hours for a next step. None of this requires new software.

Leak 4: they interview and everybody says no

This is the one that actually explains most 90 day roles, and it is the one nobody wants to look at, because the finding is that the spec is wrong.

Read the rejection reasons for every candidate who reached the hiring manager. If the reasons contradict each other, the bar is not written down anywhere and each interviewer is testing against a private version of the role. If the reasons all say the same thing and that thing is not in the job description, you have been sourcing against the wrong definition since day one. Either way, more candidates will not save you. You will just reject more of them, more slowly.

The repair is a real intake, not a job description refresh. Get the hiring manager to name the two or three things this person must be able to do in the first 90 days, the last person who did this well and what made them good, and the one requirement they would drop if it meant filling the role six weeks sooner. Our team has run this conversation across roughly a thousand placements, and the "what would you drop" question is the one that moves reqs. Write it down properly using a hiring brief built for AI sourcing, because a vague brief produces a vague pipeline no matter what tooling sits behind it.

Put a real number on what the open role is costing

Most teams treat a stale req as an annoyance rather than a line item, which is why it survives another quarter. Do the arithmetic once:

Daily cost of vacancy = (annual value the role produces, or quota, or the revenue it directly supports) divided by 260 working days. Add the hours your team is spending covering the gap, at their loaded rate. Add the recruiter and hiring manager hours already sunk into 90 days of running this req.

Whatever that daily number is, multiply it by 90. That is the sunk cost. Multiply it by 30 and that is the price of deciding to wait another month. Set it against the real cost to hire through a recruiter and the decision usually stops being a close call. The comparison people make is fee versus no fee. The comparison that matters is fee versus another 30 days of vacancy.

Where AI helps a stale role, and where it does not

AI is genuinely good at leak 1 and leak 2. Building a bigger qualified pool across a lot of sources, keeping outreach specific at a volume no human sustains, and surfacing the adjacent profiles a title based search never returns. That is real work, and it is fast.

AI does not fix leak 3 or leak 4. No tool makes your hiring manager open calendar slots, and no tool tells you that the spec is wrong, because the spec is exactly what you fed it. Point automation at a broken definition and you get more of the wrong candidates, sooner. That is not a better outcome, it is the same outcome with a shorter feedback loop.

So the sequence matters. Diagnose, fix the definition and the process, then add sourcing capacity. Doing it in the other order is how a 90 day role becomes a 150 day role with a bigger invoice attached.

If you are staring at a req that has been open a quarter and you cannot tell which of these four leaks you have, that is usually a one conversation problem rather than a tooling one. Our AI Talent Partner service does exactly this, structured intake, sourcing across a wide passive pool, personalized outreach, and pre screened shortlists, on a monthly subscription instead of a percentage of salary. Have a look at nextchaptertalent.ai if that is where you are, or send me the funnel numbers and I will tell you what I would look at first.

FAQ

How long should a role stay open before I intervene?

Run the funnel autopsy at 30 days, not 90. At 30 days you still have enough pipeline data to see the shape and enough time to change the spec before the hiring manager is frustrated. By 90 days you are usually restarting rather than adjusting.

Should I add a second agency to a role that is already 90 days old?

Only after you know which stage is leaking. If candidates are dying at the hiring manager interview, a second agency sends more people into the same rejection, and you pay for the privilege. If the top of the funnel is genuinely thin, extra sourcing capacity is a reasonable answer.

Does AI sourcing help on a role that is already stale?

Yes, if the problem is pool size or reply rate, and no if the problem is your spec or your scheduling. Fix the definition first. Automating outreach against a definition your hiring manager keeps rejecting just produces rejections faster.