Time to fill starts when a vacancy opens and stops when someone accepts the offer. Time to hire starts later, when the person you eventually hire enters the process, and stops at the same acceptance. The first measures the vacancy; the second measures one candidate’s journey. Both are speed measures, and neither tells you whether the hire was any good. That last question has its own name, quality of hire, and it is the one most dashboards skip.
This post sets the two clocks side by side, explains why a falling average can sit on top of a worsening process, and lists what we would put on the dashboard instead.
Time to hire vs time to fill, side by side
| Time to fill | Time to hire | |
|---|---|---|
| Clock starts | The vacancy opens: requisition approved or job published | The hired person applies, is sourced or is referred |
| Clock stops | Offer accepted (some teams use the start date) | Offer accepted |
| What it follows | The role | One candidate |
| Question it answers | How long did the team go without this person? | How long did our process take once the right person was in it? |
| Typical blind spot | Roles still open at the end of the period are not counted | Everything that happened before the hire showed up |
The exact start and end events vary between tools and teams, which is why our glossary entries for time to hire and time to fill each name a convention before giving a formula. Whatever you pick, write it down and keep it for the whole reporting year.
An illustrative role makes the gap visible. The ad goes live on 6 May. The person you end up hiring applies on 27 May and accepts on 10 June. Time to fill is 35 days; time to hire is 14. The 21 days in between passed before the eventual hire had applied, so they point at sourcing or advertising: the right person had not seen the ad yet, or had seen it and not applied. Read only time to hire and you would conclude the process was quick. Read only time to fill and you would blame the interviewers.
That is the practical use of the comparison. A large gap between the two points upstream, at the job ad and the channels it ran on. A small gap with a long time to hire points at the process itself: slow feedback, a third interview nobody needed, an offer stuck waiting for sign-off.
Why the average flatters a weak process
SHRM’s recruiting benchmarking reports a median time to fill of 39 calendar days for nonexecutive roles. A useful reference point, and also a reminder of how much a single number has to compress. Three problems sit inside it.
First, averages mix roles that should never be compared. A warehouse role filled in nine days and a senior engineer filled in eighty produce a mean that describes neither.
Second, both measures count only roles that closed. A team that fills four easy roles quickly and leaves two hard ones open all quarter reports an excellent number. The open roles, which are the ones costing money, are invisible.
Third, once speed becomes the target, the cheapest way to improve it is to hire whoever is available sooner. Eric Ries made the general point in Harvard Business Review in 2010: a vanity metric looks good on a slide but does not tell you what to do next. Time to hire on its own fits that description. It moves, and the movement rarely tells anyone which step to change.
What to measure instead
None of these replaces the two clocks entirely. They explain them.
- Time to first qualified interview. From the day the role goes live to the first interview with someone who meets the must-have criteria. It isolates the part you control most, sourcing and screening, from the part you control least, which is the candidate’s decision. When it runs long, look at the ad and the channels before you look at the interviews; the knockout-questions-first screening approach is one way to shorten it.
- Stage conversion. The share of candidates moving from each stage to the next: application to screen, screen to interview, final interview to offer, offer to acceptance. A low rate at one step tells you where to work. A weak offer-to-acceptance rate is a salary or expectations problem; a weak screen-to-interview rate means the screen and the hiring manager disagree about the profile.
- Decision time after the final interview. The days between the last interview and a yes or no. Candidates with other options do not wait. This number is mostly about calendars and sign-off, so it is the one a small team can fix in a week by agreeing the interview scorecard before the final round.
- Hires per source. Count hires by channel, then divide by the applications that channel produced. Applications per channel rewards the noisiest board; hires per channel shows which one delivers people you actually hire. Pair it with the retention check below to see which channel delivers people who stay.
- Quality of hire. The measure everyone agrees on and few manage. LinkedIn’s Future of Recruiting 2025 found 89% of talent acquisition professionals expect measuring quality of hire to become more important, yet only 25% feel highly confident their organisation can do it. A workable proxy for a small team: of the people hired a year ago, how many are still here, and would their manager hire them again? Check the cohort at 90 days and again at twelve months. Work Institute reports that over one-third of new hires quit within their first year, so the first-year cohort is where a weak selection shows up first.
The first three move while a role is still open, so you can act on them this week. The last two are slower and confirm, months later, whether the faster numbers were telling the truth.
Measuring it with the data you already have
You do not need a BI project. You need the same few dates recorded for every application, the same definitions all year, and a quarterly look.
Record, per application: date applied, source, the date of each stage move, date of the final interview, decision date, offer date and acceptance date. Per role: date opened and date filled. Report each measure as a median together with the count of hires behind it, and list the roles still open next to it so the closed-only bias stays visible.
A disclosure, since we sell an ATS: Join’s own recruiting analytics dashboard is still in development, so here is what the product records today and what you would still do by hand. A job’s Postings tab shows how many candidates and clicks each job board produced, which covers the applications half of hires per source. The hires half is a count you do by hand: each application on the candidate profile shows its source and date (for example, applied via LinkedIn), and the profile’s activity log time-stamps every stage move, which is where decision time comes from. On the Advanced and Enterprise plans you can filter applications by job, job status or stage and export them to CSV or XLSX, one row per application, as the base for your own spreadsheet. Retention you will have to take from HR records, since Join does not see who leaves in month eight.
What most teams get wrong
They redefine the clock halfway through the year. Moving the end event from offer accepted to start date adds the new hire’s notice period to every role overnight, and the trend line now compares two different measures.
They report one company-wide number. Split by role family at minimum, or the mix of roles hired that quarter decides the result.
They set a target on speed and nothing else. A team chasing a lower time to hire will cut the reference call and the second opinion first, which are the steps that protect quality of hire. Keep time to fill and time to hire on the dashboard as context. Put the decisions on the measures that explain them.
For how to pick the full set of measures and test whether your data can support them, see our research on recruiting KPIs.
