Time to hire: where the weeks actually go

Three different numbers get called roughly the same thing, and teams argue past each other because of it.

• Time to fill runs from the role being approved or advertised to somebody accepting. It measures your whole pipeline, including the stretch before anybody applied.
• Time to hire runs from a candidate applying to that candidate accepting. It measures your process rather than your reach.
• Time in process is the same span for every candidate, including the ones you rejected. It is what candidates experience, and it is the number that loses you people.

They move independently. A role can have a quick time to hire and a dreadful time to fill because nobody applied for a month. Better adverts and better reach fix one; a better process fixes the other, and confusing them means putting the effort into the wrong half.

Count calendar days, not working days. A candidate waiting over a weekend is waiting.

Here is a first-round-plus-panel process for one candidate, with what sets each stage's pace, and whether anybody is doing anything during it.

Swipe the table sideways to see every column.

No durations are given, because the honest ones vary enormously by team and no average worth quoting could be sourced. What does not vary is the shape: seven of those ten rows are waiting, and during them nobody is doing anything. The interviews, which is what teams try hardest to compress, are the short rows.

One booking is a small thing. A hiring process is dozens of them, in series.

Each needs a candidate who mostly cannot talk during working hours, an interviewer whose calendar is already full, and often a time zone in between. The proposal goes out, the reply arrives the next day, the slot has gone, a new proposal goes out. Then somebody is ill and it happens again.

Panels are worse, because they batch. A slot four people share is scarcer than any one of their calendars suggests, and that single constraint is frequently the largest block of elapsed time in a hire.

The second-order cost is quieter. Scheduling load caps how many people you speak to at all: nobody books forty phone screens, so the pile gets cut on documents to keep the calendar survivable. That is how a scheduling problem becomes a quality problem without anyone deciding it should.

• Approval to open the role. Rarely counted, because the clock usually starts at the advert.
• Take-home turnaround. A weekend task adds a round trip: the candidate needs their weekend, then somebody needs a slot to review it.
• The debrief that waits for a meeting. If decisions happen in a weekly meeting, every decision costs up to a week no matter how quickly it was made.
• Offer sign-off. Compensation approval, a contract template somebody has to edit, a signature from a person who is travelling.
• The candidate's own timeline. They may be waiting to compare offers, or on a notice period nothing you do can shorten.
• Legal lead time, where it applies. If you hire in New York City and use an automated employment decision tool, Local Law 144 requires notice to candidates before it is used, which is lead time to plan for rather than discover. Ask counsel what applies to you.

Added up, these are often more elapsed days than the interviews themselves.

Anything that removes a wait shortens a hire. Anything that removes a step usually just relocates the risk.

1. Take the first round off the calendar. The largest structural win available, because it deletes one booking per candidate and all the waiting around it.
2. Set the decision rule before you start. Who decides, on what evidence, by when. Most debrief delay is not analysis; it is nobody being certain they are allowed to decide.
3. Book the panel before you need it. Hold a recurring slot with the same interviewers and fill it, instead of hunting a fresh slot per candidate.
4. Pre-approve the offer band with the role, not with the candidate in hand.
5. Reject quickly. It does not shorten the hire you made, but it is the part candidates notice, and it returns your own attention to the shortlist.
6. Instrument the waiting rather than the interviewing, which is the next section.

• Cutting a round. That removes information, not delay. If the round existed only out of habit, delete it; if it tested something, replace the test.
• Dropping structure to move quicker. Unstructured conversations are quicker to arrange and far worse to compare. The re-analysis by Sackett, Zhang, Berry and Lievens found structured interviews top-ranked among the selection procedures it examined, and structure is exactly what gets abandoned when a process is rushed.
• Offers with a deadline attached. Compressing the candidate's decision poisons the start of the relationship, and the strongest candidates are the likeliest to walk.
• Optimising an average. A mean hides the role that took four months. Work on your slowest hire, not your typical one.
• Buying a scheduling tool for a process with no decision rule. A faster booking into a debrief nobody owns saves a day and loses a week.

One number for a whole hire tells you nothing you can act on. Record four timestamps per candidate and the bottleneck is obvious within a fortnight.

1. Applied.
2. First round complete.
3. Decision made on the first round.
4. Offer sent, and accepted or declined.

That gives three intervals: how long applicants wait to be interviewed, how long a finished interview waits for a decision, and how long the offer takes. Whichever is largest is the only one worth working on, and it is seldom the one people assume.

There is deliberately no benchmark here to compare yourself against. Published averages vary by role, market and counting method, and we could not open a primary source we were willing to quote. Your own four timestamps are worth more than somebody else's mean.

Career1 removes one row of the table above: the row that repeats for every candidate, booking and holding the first-round interview.

Applying to a job starts it. The AI recruiter interviews each applicant by voice in the browser for about eight minutes, at whatever hour they choose, with no slot negotiated with anybody. The report, transcript and recording are waiting when your team next looks, so the interval between somebody applying and you having real evidence about them is however long they took to sit down, not however long two calendars needed to agree. A sample report shows what arrives.

What it does not shorten, and does not claim to: the stretch before anybody applies, your internal decision, the panel, offer approval, a notice period, or a candidate weighing somebody else's offer. Career1 schedules nothing with humans and chases nobody. If your bottleneck is a debrief nobody owns, this saves days and leaves the weeks exactly where they were. The honest version of the promise is that you stop doing the first-round screening, not that hiring becomes instant. Plans priced by monthly interview volume are on the pricing page.

1. Create a workspace: an email address and a password, no card.
2. Post the role you are slowest on.
3. Share the job link, and set the decision rule now: who reads the reports, and by when.
4. Applicants interview themselves, at any hour. Nothing gets booked.
5. Read the reports at the top of the ranking and move people straight into a human round.

The free plan vets two candidates a month and needs no card. If the pile is large, the process for that is in screening 200+ applicants, and if you want to price the hours you spend today, the arithmetic is in what screening really costs.

Both sources below were opened and checked on 26 September 2026.

• Sackett, Zhang, Berry and Lievens, "Revisiting meta-analytic estimates of validity in personnel selection", Journal of Applied Psychology: doi.org/10.1037/apl0000994.
• New York City guidance on automated employment decision tools, including the candidate notice: nyc.gov.

No average time to hire or time to fill appears anywhere above, because no primary source for one could be opened.