Time to fill and time to hire can end in the same event and still answer different questions. The most common mistake is to call any draft duration by one of these names without recording where the count began. When this happens, two dashboards display similar numbers, but not necessarily comparable.
In the SHRM glossary, time to fill measures the interval from the vacancy request to the acceptance of the job, while time to hire measures the time between the entry of the professional who will be hired into the pipeline and the acceptance of the offer. The final milestone is similar; the starting point changes. This difference is enough to change the reading about where the process is taking.
The difference is in the starting point
For a clear operational measurement, treat time to fill as the time between the approved request and the acceptance of the offer. Treat time to hire as the time between the hired candidate entering the pipeline for that vacancy and acceptance of the offer. In simple notation: time to fill = date of acceptance - date of approved request; time to hire = date of acceptance - date of entry of the contractor into the pipeline.
Choosing the approved request as the starting point makes it clear when the organization actually authorized the search. If your company starts counting on the creation of the requisition, the opening of the vacancy or the publication of the advertisement, this may be valid for an internal indicator, but the name and definition must appear on the panel. Changing the milestone without renaming the metric destroys the historical series.
The same goes for the ending. Offer sent and offer accepted are different events. A proposal can remain under negotiation for days. If the goal is to follow the definition that ends in acceptance, using the offer creation or submission date artificially shortens the interval.
The dictionary of events that the ATS needs to record
The safest way to avoid confusion is to build a small dictionary of events before building the formula. The ATS can have dozens of fields, but four dates solve most of the problem: requisition approved, candidate enters pipeline, offer accepted, and hire starts.
Request approved
This is the moment when the vacancy stops being just an intention and becomes authorized for recruitment, according to the adopted internal rule. Do not confuse it with the creation of a draft, with the date on which the manager informally asked for the position or, necessarily, with external publication. If financial approval, ATS opening and publication happen on different days, the dictionary should say which one starts the time to fill.
When the system does not have a reliable approval field, it is better to explicitly assume another milestone, such as opening date, than to retroactively fill in an estimated date. The indicator remains useful, but now measures openness until acceptance, not approval until acceptance.
Candidate enters the pipeline
For time to hire, the beginning must be specific to the person who was hired and the vacancy in question. For a candidate who has applied, this could be the application date. For someone actively approached, this could be the date they were matched with the position and entered the process. The important thing is not to use the date on which the profile was created in the talent pool months before, if it was not yet being considered for that request.
This detail matters in reactivations. A professional can exist in the ATS since 2024 and enter the pipeline for a vacancy in 2026 only when the recruiter links him to the process. Measuring from the original profile creation would transform time to hire into a mixture of relationship time and selection time.
Offer accepted
Record the actual date on which the candidate accepted the proposal. If the recruiter updates the ATS two days later, the system should preserve, when possible, the effective date of acceptance, not just the moment of editing. This difference prevents administrative delays from appearing as slow selection times.
Greenhouse's documentation reinforces the importance of maintaining correct application, offer, acceptance and vacancy opening dates so that hiring reports are not contaminated by late or incorrect records. The lesson applies to any ATS: the quality of the metric depends on the quality of the event.
Start of contract
Start date is important for capacity planning, but should not replace acceptance if you are calculating time to fill or time to hire with the above definitions. Between acceptance and the first day there may be prior notice, exams, documentation, checks and personal availability. This interval measures something else.
If the company needs to monitor this, create a third indicator, for example accept until the beginning. Therefore, a 30-day wait to begin does not artificially inflate recruitment speed.
The same fictitious process produces two metrics
Imagine a vacancy whose request was approved on March 3, 2026. The vacancy was published on March 5. The candidate who was eventually hired entered the pipeline on March 12th. The offer was sent on March 27th, accepted on March 31st, and commencement took place on April 20th.
With a count of elapsed days, the time to fill is March 31st minus March 3rd: 28 calendar days. The time to hire is March 31st minus March 12th: 19 calendar days. The interval between approved request and start is 48 days, but this number is not one of the two metrics defined here.
This example also shows why time to fill usually captures a larger part of the journey. The nine days between approval of the requisition and the hired candidate entering the pipeline appear in the time to fill, but not in the time to hire. They may reflect job preparation, posting, sourcing, or simply the time it took to find who would be hired.
Calendar days or business days: choose a convention and don't change in the middle
Workable describes time to fill on calendar days and also warns that companies can choose different starting points, as long as the rule is consistent. For external comparisons, calendar days tend to be the easiest choice to audit because weekends and holidays are not dependent on each area's calendar. For internal SLAs, business days may make sense, but should appear as a separate or clearly labeled metric.
In the March example, counting only Monday to Friday and without taking into account additional holidays, the 28 calendar days of time to fill become 20 working days; the 19 calendar days of time to hire turned into 13 working days. The two numbers describe the same process, but they cannot be placed in the same series as if they were equivalent.
Also define whether the count is of elapsed days. A simple and reproducible convention is to calculate end date minus start date, so that a process started and ended on the same date has zero elapsed days. If the company prefers to count the first day as day 1, it can do so, but it must maintain this rule in all vacancies and in all versions of the panel.
For business days, use a formal corporate calendar. It is not enough to remove Saturdays and Sundays if the organization intends to exclude national, local holidays, or internal dates. The calendar used must be part of the technical definition of the indicator.
The dashboard label is not enough
Even ATS products use similar names for different calculations. Greenhouse documentation updated in 2026 shows that its core Time to Hire report measures, in one of its views, from the application date to the job creation date, whereas a dashboard might use the hire date. The Time to Fill by Job report uses the opening date through the date the person is marked as hired.
This does not mean that the software is wrong. It means that the name displayed in the interface does not replace the company's data dictionary. If you want to measure pipeline until acceptance, configure the formula with these events or export the necessary data to calculate it. If you decide to use the native ATS metric, document exactly what it does.
The best practice is to show a short line with the definition on the panel itself. Example in text: Time to hire = entry of the contractor into the pipeline until acceptance, calendar days. This sentence is worth more for comparability than a sophisticated graph without an explicit rule.
Mean and median tell different stories
After defining the dates, a second risk arises: summarizing heterogeneous processes with a single average. The average is simple and useful for tracking total time consumed, but it reacts strongly to extreme cases. The median shows the center point of the distribution and tends to better represent the typical experience when there are a few very long vacancies.
If five hires took 21, 23, 24, 26 and 68 days, the average is 32.4 days and the median is 24 days. Neither number is wrong. The average reveals that the 68-day case led to the overall result; the median shows that half of the processes took around 24 days or less and half took around 24 days or more.
Therefore, a useful dashboard can display the average, median and number of cases in the same period. With enough volume, percentiles like P75 also help answer how long most hires take without letting a single outlier dominate the reading.
Segment before comparing areas or periods
Mixing all positions tends to produce a metric that is apparently precise and not very actionable. A senior leadership position, a high-volume operational position, and a rare technical role have different markets, stages, and approvals. If the job mix changes from month to month, the overall average may rise even as each segment improves.
Start with segmentations that reflect real process differences: job family, seniority, location, business unit and hiring model. Candidate and recruiter sourcing can also be useful for specific questions, but require care not to attribute causality to groups with different job profiles.
Always show the number of cases. An 18-day median based on two openings should not be given the same weight as a 24-day median based on one hundred hires. On small teams, quarterly or semi-annual windows may be more stable than monthly cuts.
Set up the measurement reproducibly
Choose the unit of analysis
Define whether each line represents a requisition, an individual opening or a contract. This is especially important in requests with multiple positions. If a single requisition provides for five hires, the first acceptance and the fifth acceptance represent different moments. A company can measure each opening separately or measure the requisition until all vacancies are filled, but it should not switch between the two approaches without warning.
For time to hire, the natural unit is usually the completed hiring, because the start depends on the candidate actually hired. For time to fill, the unit can be the individual opening when the ATS supports multiple openings within the same request.
Define which event places the case in the period
A monthly dashboard also needs to say how the cohort is formed. For completed duration metrics, a simple option is to include hires whose acceptance occurred in the month. Thus, each case has a known beginning and end, even if it began a month earlier.
Vacancies still open must be monitored using another indicator, such as requisition aging, calculated from
approval to date. Mixing open cases with completed times requires different techniques and can skew the average because the longest cases have not yet finished.
Preserve original events
Do not overwrite the application date to correct a change in stage, nor reuse the same field for both the submitted offer and the accepted offer. Ideally, you should store event timestamps and derive metrics from them. If a manual correction is necessary, record who changed it, when they changed it, and what the previous value was.
When the indicator rule changes, version the definition. For example: Time to fill v1 = opening date until hire date; Time to fill v2 = request approved until accepted. The old historical series can continue to exist, but it should not be glued to the new one without recalculating the previous periods with the same logic.
Audite anomalies before publishing the number
Create simple quality tests. Negative duration indicates reversed dates or wrong association. Zero days in high volume may signal a late ATS update. Acceptance prior to entry into the pipeline is impossible under the chosen definition and needs review. Requests without a start date should not silently enter the average.
Also look for outliers before discarding them. A 180-day vacancy could be a registration error, but it could be a real process. Fixing a bug is different from removing a hard case just because it makes the indicator worse.
What each metric helps diagnose
Time to fill is more useful for planning because it starts before the hired candidate's trajectory. It helps answer how long the organization takes from authorizing the need to ensuring someone has accepted the position. If it is high, the delay may be in approval, opening, sourcing, screening, interviews, decision or negotiation.
Time to hire cuts the speed from the moment the hire entered the pipeline. It is more sensitive to the efficiency of screening and moving that candidate through the process. If time to fill is high and time to hire is low, one hypothesis is that a large part of the delay is before the hire enters the pipeline. If both are high, it is worth investigating further steps, but the metric alone does not prove the cause.
This reading becomes stronger when the intervals are broken down by stage. Instead of just asking why the time to hire increased, check how long the candidate spent between application and screening, screening and interview, interview and decision, decision and offer, offer and acceptance. The total indicates that there is a problem; the gaps in between help you locate it.
The number is only comparable when the rule travels together
Before bringing time to hire or time to fill to a meeting, make the number carry its own definition. The panel must inform the initial event, the final event, whether the count uses calendar or working days, which event defines the cohort for the period, the unit of analysis and how many cases make up the result.
For external comparison, first confirm that the other source uses the same landmarks. A 40-day benchmark based on publish to accepted is not directly comparable to an internal indicator of request approved to start. The difference may only reflect the formula.
The separation between the two metrics is simple when the ATS preserves the right events. Request approved until accepted responds to how long the vacancy took to be filled within the chosen convention. Entry of the contractor into the pipeline until acceptance responds to how long it took to convert that candidate into a hire. The rest is data governance: recording the milestones, keeping the rule stable, and comparing equivalent groups.
