Updated 1 November 2026
Time to hire, in numbers
A percentage means little until you know what it is a percentage of
Ask how much AI interviews cut time to hire and most pages answer with one percentage. Few say what it is a percentage of, who measured it, or which part of hiring got faster.
Fabric, an AI interviewer that screens candidates by phone call and live video interview, does cut time to hire. When first calls and interviews wait on recruiter calendars, it cuts a lot. How much it cuts for your team depends on three things: your baseline, how your numbers define time to hire, and which stage gets faster.
Pin those three down and the percentage becomes arithmetic you can check. The calculator below runs it on your own numbers.
Time to hire vs time to fill: what each one measures
Each source starts and stops the clock in a different place, so two honest numbers can differ.
- Time to fill, as SHRM measures it: days from the requisition being opened to the candidate accepting the offer, in calendar days.
- Time to hire, as The Josh Bersin Company measures it: days from the requisition being approved to the candidate accepting.
- Time to screen: days from an application to a finished screening call or interview. Zapier measures it from the first review of an applicant.
Vendor claims often use none of these. Chipotle's widely quoted result runs from application to start date, which includes onboarding. Most buyers would not call that time to hire.
The same result shows why the arithmetic matters. Paradox's Chipotle case study reports 12 days falling to 4. That is a 67% cut, yet the headline says 75% and the page does not explain the gap.
This page uses "time to hire" for the whole span from requisition to accepted offer. It says "time to screen" when it means the screening wait alone.
What is a normal time to hire in 2026?
About a month and a half. SHRM's 2026 benchmark puts the median time to fill at 39 calendar days for nonexecutive roles. Larger organisations take longer, and each figure in the table answers a slightly different question.
| Figure | What it measures | Source and sample |
|---|---|---|
| 39 days | Median time to fill, nonexecutive roles (requisition opened to offer accepted) | SHRM 2026 benchmarking, data from 4,600+ organisations |
| 45 days | Median time to fill, executive roles, down from 60 days | SHRM 2025 Recruiting Benchmarking, 2,371 US HR members, surveyed 9 January to 3 March 2025 |
| 61 days | Median time to fill, nonexecutive roles, at organisations with 5,000+ employees | Same SHRM 2025 survey |
| 44 days | Average time to hire (requisition approved to offer accepted) | The Josh Bersin Company with AMS, report of 1 June 2023, reported by HR Dive. The full report is not public |
Seven different "industry averages" circulate on vendor pages: 36, 39, 41, 42, 44, 45 and 46 days. Three are credited to SHRM. They differ because they cover different spans, different organisations and different years.
A cut quoted against an unnamed baseline tells you nothing. Half of 25 days and half of 61 days are very different savings. SHRM's full reports are for members, so every figure above comes from a public page.
Where do the days go in hiring?
No public source says how many days each stage takes. SHRM splits time to fill into seven stages. The evidence that exists points to approvals, interview scheduling and offers as places days pile up. Fabric works on the stretch in the middle.
The list shows SHRM's seven stages in order and marks the ones Fabric works on.
SHRM's seven stages of time to fill, and where Fabric works
- Requisition opened to approvedThe approval waitYour team
- Approved to job postedThe posting waitYour team
- Job posted to screening startedFabric calls within 24 hours of the job closingFabric, from the close
- Screening all applicantsCalls and interviews, up to 500 at onceFabric
- Conducting interviewsFabric runs the first round and books the later onesFabric, first round
- Final decision and offer sentYour team decidesYour team
- Offer to acceptanceThe candidate decidesThe candidate
Purple stages are the ones Fabric works on. The others sit with your team and your candidates.
The table sets the evidence for each stage beside whether Fabric works there.
| Stage | What the data shows | Does Fabric work here? |
|---|---|---|
| Approval and posting | SHRM 2025 (2,371 US HR members): organisations with 5,000+ employees take nearly twice as long as the median to get positions approved and jobs posted. | No. It happens before applicants arrive. |
| Screening | No independent benchmark names it the longest stage. In Zapier's own account of its pilot of another vendor's AI interviews, about 250 interviews, the time from first applicant review to a completed screen fell from 8 days to 2.75, a 66% cut. The AI interview largely replaced the recruiter's live screen. | Yes. Fabric calls shortlisted candidates within 24 hours of the job closing and runs interviews around the clock. |
| Interview scheduling | GoodTime, which sells scheduling software, surveyed 500+ US talent leaders: recruiters spend 38% of their time scheduling interviews. Ashby, a recruiting software company, studied 2.8M+ candidates: about 5 to 6 days from asking for availability to the interview starting. | Partly. The first round needs no calendar slot, and Fabric books the later rounds. |
| Panel availability | Ashby: 34% of panel interviews were scheduled within a day, against 54% of single interviews (excluding direct booking). SHRM 2025: organisations with 5,000+ employees also take nearly twice the median to conduct interviews. | No. It depends on your interviewers' calendars. |
| Offer | Gartner, nearly 3,000 candidates, January to March 2025: 35% backed out after accepting an offer, and 44% had more than one offer. | No. It depends on pay, growth and your close. |
Both scheduling sources sell software, so weigh them accordingly. Their samples are disclosed, which most vendor figures are not. Not sure whether your delay is booking or a shortage of interviewer hours? Fabric's guide to AI interviewers vs interview scheduling software shows how to tell.
The table has a plain consequence. A screening tool can only shorten the stretch it runs. Zapier's pilot shows how far that stretch shrank at one company where candidates queued for a recruiter. Your own gain is the size of your queue, and the calculator below measures it.
What AI interviews cut, and how much for you
An AI interviewer takes the waiting out of the screening stretch. Three things change there.
The first call stops waiting for a recruiter's calendar. Fabric's AI calls candidates shortlisted by resume screening within 24 hours of the job closing. It retries three times over two days, so a missed call is tried again within two days. Fabric's clock starts when the job closes, so the time a job stays open is not changed.
Screening runs in parallel and around the clock. Fabric can hold 500 interviews at once. The AI interview is a live video conversation that works in a phone browser with no app, even on weak 4G.
Results arrive as a scorecard instead of a recording to re-watch. Recruiters can review every result and decide who goes through.
You can size the saving with plain arithmetic, the same way Fabric's guide to bulk hiring drives does. Indeed puts a phone screen at 15 to 30 minutes. Screening 5,000 candidates by phone is 1,250 to 2,500 recruiter hours before any scheduling, notes or follow-ups.
Now do the same for your team. Every starting value below is an example, so change each one to match your hiring.
Your numbers
Work out your own time to hire
Plain subtraction. Nothing here is a Fabric result.
Recruiter time on screening calls today: 100 hours.
It assumes every other stage stays as it is. Fabric starts calls within 24 hours of the job closing and retries over two days. Applicants who applied while the job was open also wait for it to close, so replace the 3 days with your own pilot data once you have it.
Try a team with a backlog: 45 days in total, 20 of them waiting to be screened, down to 3. That is 17 days off 45, or 38%. A team that already screens within 3 days has nothing to take back. That spread is why one percentage cannot fit everyone.
Format decides whether candidates finish
Candidates finish a live video call far more often than a one-way recording. A pre-registered field experiment by Avery, Ip, Leibbrandt and Vecci (working paper, March 2026) randomised more than 3,000 real applicants to jobs at tech companies. Figures below are approximate.
Share of applicants who kept going
Pre-registered field experiment, more than 3,000 real applicants, March 2026. Figures are directional.
Telling applicants that an AI, not a person, would score the recording changed completion by only about 3 percentage points. That gap was small enough to be down to chance. The one-way format drove the drop-off, not the AI label.
This matters for time to hire. If most invited candidates drop out, a team has to find more, and that costs days. Fabric's interview is a two-way conversation, not a recording. It has AI disclosure and recording consent built in, and a candidate whose connection drops rejoins where they left off. The experiment did not test Fabric's interview, so it gives no completion rate for Fabric.
Whichever tool you use, count how many invited candidates finish, and say what you divided by.
Five questions to ask of any "cut time to hire" claim
Use these on a vendor page, a case study or this one.
- Against what baseline? Seven "industry averages" circulate, from 36 to 46 days. A cut with no baseline cannot be checked.
- Which metric? Time to hire, time to fill, time to screen, or a span that includes onboarding. Each gives a different percentage for the same hiring.
- One customer or many? A "customer average" can be one customer's quote, restated as if it were typical.
- Completion of what? A vendor's completion rate can look far higher than the experiment above because one divides by the candidates who started and the other by everyone invited.
- What year and what sample? A figure with no year and no number of people behind it is a slogan.
Even a claim that survives all five describes someone else's hiring. Your own numbers decide.
What to measure instead of one headline number
Measure the stretch between each pair of events, not only the total. If you track only the total, you cannot tell whether a tool worked or something else moved. A new approval step can hide a faster screen.
Four timings that show whether a tool worked
Add two ratios: completion with its denominator stated, and the share of screened candidates who pass the next stage.
Take these timings for a few months before you change anything. Then you can say which stretch got shorter and by how many days.
Frequently asked questions
How much do AI interviews reduce time to hire?
There is no single number, and the 30% to 75% figures on vendor pages are mostly unsourced. AI interviews shorten the wait to be screened, so your total falls by the days you take out of that wait: the calculator above subtracts your screening wait today from the wait you expect.
Do AI interviews actually save time, or just move the work?
They save time on the screening wait and move some work to review, because recruiters read a scorecard instead of sitting on every call. The saving is largest when the AI screen replaces the recruiter's own first call, not when it is added on top.
How long does AI interview software take to implement?
Fabric's setup takes about 30 minutes, and ATS integration is on the Scale and Enterprise plans. After that, a recruiter enters a job description and Fabric's AI agents take it from there.
What Fabric leaves alone
Fabric does not change how long approval takes, how long a job stays open before it closes, whether your interviewers are free for panels, or whether a candidate accepts the offer. Your team still makes the final hiring decision. That is why the calculator subtracts only the screening stretch. Read more about the AI interviewer. If you are weighing tools, the AI interviewer buyer's guide covers how to evaluate them.