Time-to-Hire Is a Queueing Problem, Not a Sourcing Problem

We cut average time-to-hire from 41 days to 19 across eleven open roles. We did not add a sourcer, buy a new tool, or increase the recruiting budget. We measured where candidates were waiting and deleted the waiting.

Almost every team I meet treats slow hiring as a supply problem. Not enough candidates, not enough sourcing hours, not enough job board spend. So they buy more top-of-funnel. Then time-to-hire stays flat, the pipeline gets noisier, and the recruiters get busier without getting faster.

Here is the thing nobody wants to hear: in a typical pipeline, a candidate spends most of their time doing nothing. They are not being assessed. They are sitting in a queue, waiting for a human on your side to act. If your process takes 41 days, a third or more of those days are probably dead air.

Forty-one days is not an unusual starting point, which is part of the problem: it feels normal. SHRM's 2025 recruiting benchmark, a survey of 2,371 HR professionals, put median time-to-fill at about a month and a half, and its 2026 update reported a median of 39 days for non-executive roles. (SHRM measures from the day the requisition opens to the day the offer is accepted, which is the definition I use here too.) Average is not the same as fine.

Why more candidates makes it slower

There is a sixty-year-old result in operations research that explains why buying more candidates does not help. Little's Law, proved by John Little in 1961, says that in any stable system the average amount of work in progress equals the rate at which work is completed multiplied by the average time each item spends in the system. Rearranged, which is how factories use it: cycle time equals work in progress divided by throughput.

Translate that into hiring. Work in progress is the number of candidates in your pipeline. Throughput is how many your team can actually move through a stage each week, which is set almost entirely by interviewer hours. If those hours are fixed and you add candidates, the only thing that can give is the time each candidate waits. More sourcing does not shorten the line. It lengthens it.

So the levers are throughput (more protected interviewer capacity) and the waits between steps. That is where the rest of this post goes.

Measure the gaps, not the stages

Most applicant tracking systems report stage duration: how long someone sat in "Phone Screen" or "Onsite". That number blends two very different things: the time the work took, and the time before anyone started the work.

Split them. For every candidate, record two timestamps per stage: when they became eligible for the stage, and when the stage actually happened. The difference is queue time. That is the number to attack.

When we did this for the first time, the result was uncomfortable:

• Each interview itself: 45 minutes
• Time from "ready to schedule" to "interview happened": 6.4 days
• Time from "interview happened" to "written feedback submitted": 3.1 days
• Time from "final interview" to "offer approved": 5.7 days

Under an hour of actual assessment per stage. Over fifteen days of waiting across just three handoffs, before counting anything lost at intake. No amount of sourcing fixes that.

The four delays

Once you measure queue time, the same four bottlenecks show up in almost every organisation.

1. The intake delay

The role is approved but nobody has agreed what "good" means. The recruiter writes a job description from an old template, sources against a guess, and the hiring manager rejects the first eight profiles for reasons that were never written down. Two weeks are gone and the pipeline restarts.

2. The scheduling delay

This is the biggest and the dumbest. A strong candidate finishes a screen on Tuesday and cannot meet the panel until the following Thursday, because three senior people have no shared availability and a coordinator is negotiating over email.

3. The decision delay

The interview happened. The feedback is in someone's head, not in the system. The debrief is scheduled for whenever the calendar allows. The candidate hears nothing for a week and quietly re-engages with the company that moved faster.

Candidates say as much when asked. In a 2021 Robert Half survey of more than 1,000 US workers, 62 percent said they lose interest in a job if they do not hear back within two weeks of the first interview, rising to 77 percent after three weeks with no update. That is a survey of stated intentions, so treat it as direction rather than precision. The stronger evidence is behavioural, and it comes later in this post.

4. The approval delay

Everyone agrees on the hire. Now compensation has to be modelled, a band has to be confirmed, and two executives have to sign. None of this required the candidate to be present, and all of it could have happened before the candidate ever applied.

What we changed

Four mechanics, each aimed at one queue. None of them are clever. All of them are enforced.

A real intake meeting, before the role opens. Forty-five minutes with the hiring manager. We leave with a scorecard: four to six outcomes the person must deliver in their first year, the two or three signals we will assess each one with, and one explicit trade-off (what we are willing to be weak on). If we cannot name what we will compromise on, the role is not defined yet. We do not post it.

The scorecard does more than save time. It is what turns a conversation into a structured interview, and structure is the single biggest lever on interview accuracy. The most thorough recent review of selection research, Sackett, Zhang, Berry and Lievens (2022) in the Journal of Applied Psychology, found structured interviews to be the top-ranked selection method, with an average validity of .42. Unstructured interviews came in at .19, less than half. (SIOP has a readable summary.) The intake meeting is where the structure gets written.

Pre-blocked interview slots. Every interviewer holds two fixed hours a week for hiring. The slots exist whether or not there is a candidate. They are protected like a customer meeting. This single change took scheduling from 6.4 days to 1.8, because coordinators stopped negotiating and started booking.

A 24-hour feedback rule with teeth. Written feedback is due within one working day of the interview, submitted before the interviewer sees anyone else's notes. If feedback is late, the interviewer loses their next slot and someone else takes the panel seat. Nobody enjoys this rule for the first month. Everyone defends it after.

Compensation approved at intake. The band, the equity range and the signing flexibility are agreed when the role opens, not when the offer is drafted. The hiring manager can extend a verbal offer inside the band without a second approval cycle. Approval delay went from 5.7 days to under one.

What did not work

I have tried plenty of things that sounded good and changed nothing.

Adding more interview stages to "increase confidence". In our data every extra stage added four to six days and almost never changed the decision. We removed two rounds and hire quality did not move.

We are not the only ones to find this. When Google analysed its own interview data, it found four interviews were enough to predict its hiring decision with 86 percent confidence, with each additional interviewer adding very little. Its "Rule of Four" cut average time-to-hire by about two weeks. (Note what that measures: agreement with the final decision, not later job performance.) Glassdoor's analysis of more than 340,000 interview reviews found that panel interviews, candidate presentations, skills tests and background checks each had a statistically significant effect in lengthening hiring time. Every stage has a price in days. Make each one earn it.

Asking people to "prioritise hiring" in a meeting. In my experience intent decays in about nine days. Calendar blocks do not.

Buying a scheduling tool before fixing availability. The tool surfaced the problem faster; it did not create hours that senior engineers had not committed.

Chasing a lower time-to-hire as a target in itself. The moment it became a number people were graded on, someone started closing requisitions early and reopening them. Measure it, publish it, but do not bonus on it. Social scientists have a name for this. The psychologist Donald Campbell put it in 1979: the more a number is used to make decisions, the more it is subject to corruption pressures, and the more it distorts the very process it was meant to monitor. The anthropologist Marilyn Strathern's shorter version, usually credited to Goodhart, is that when a measure becomes a target, it ceases to be a good measure.

The trade-off worth naming

Speed can absolutely damage quality, if you get it by cutting assessment. That is mostly not what happened here. Each remaining interview kept its full forty-five minutes, and the two rounds we removed were the ones that, on our own records, almost never changed a decision. Most of the gain came from the waiting between interviews, not the interviews themselves.

There is peer-reviewed evidence that this kind of speed is free. Becker, Connolly and Slaughter (2010) studied 3,012 candidates at one company in Personnel Psychology and found that candidates were more likely to accept earlier offers, and that people hired after quicker offers performed and stayed just as well as those hired after slower ones. Waiting cost the company acceptances and bought it nothing.

There is one real cost. Pre-blocked slots and a 24-hour feedback rule take genuine calendar capacity away from senior people. In a 60-person company, that is roughly 25 to 30 hours a week of engineering and leadership time reserved for hiring. If leadership is not willing to fund that, no process design will save you, and you should stop pretending hiring is a priority.

Start here

If you do nothing else this quarter, do these three things in this order:

1. Instrument queue time for one month. Two timestamps per stage. A spreadsheet is enough. Do not change anything yet. Just look. The number will make the argument for you.
2. Pre-block interview slots for your three busiest interviewers. Two hours each, recurring, protected. This is usually worth four to five days on its own.
3. Set the 24-hour feedback rule and enforce it once, visibly. The first time you move someone off a panel for missing it, the rule becomes real.

Then re-measure. In my experience you will find between 30 and 50 percent of your cycle time was queueing, and most of it will come back within six weeks.

The best candidates are usually in process with other companies at the same time. They are not comparing your compensation philosophy. They are noticing who respected their time. Speed is not a shortcut around rigour. It is the most visible evidence that you have any.

Sources

1. Little, J. D. C. (2011). Little's Law as Viewed on Its 50th Anniversary. Operations Research 59(3), 536–549. The original proof: Little, J. D. C. (1961), A Proof for the Queuing Formula L = λW, Operations Research 9(3), 383–387.
2. SHRM (2025). 2025 Recruiting Executives Benchmarking: Insights to Maximize Recruitment; and Recruiting benchmarking (2026 update).
3. Robert Half (10 February 2021). How to lose a candidate in 10 business days. Press release on a survey of US workers.
4. Sackett, P. R., Zhang, C., Berry, C. M. and Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology 107(11), 2040–2068. Summary: SIOP, Is cognitive ability the best predictor of job performance?
5. Shaper, S. (2017). How many interviews does it take to hire a Googler? Google re:Work (archived copy).
6. Chamberlain, A. (2015). Why Is Hiring Taking Longer? New Insights from Glassdoor Data. Glassdoor Economic Research. Based on 2009–2014 data.
7. Campbell, D. T. (1979). Assessing the impact of planned social change. Evaluation and Program Planning 2(1), 67–90 (reprint).
8. Strathern, M. (1997). 'Improving ratings': audit in the British University system. European Review 5(3), 305–321.
9. Becker, W. J., Connolly, T. and Slaughter, J. E. (2010). The effect of job offer timing on offer acceptance, performance, and turnover. Personnel Psychology 63(1), 223–241.