How to Define Time-to-Fill Consistently

To define time-to-fill consistently, write down its start event, end event, clock, pause rule, cohort, source and the treatment of incomplete or corrected records, then compare only results that use the same definition. Time-to-fill is the elapsed time between a requisition start event and a fill event. This guide is for a hiring manager waiting on debriefs.

What goes in a time-to-fill metric contract?

A time-to-fill metric contract is one sentence that another person can rerun and get the same result. Fill in this template: For [unit] in [scope], time-to-fill is [end] minus [start], measured in [clock], with [pause rule], for [mature cohort] from [source], reported as [statistic]. Keep owner, field map, process version and change rule beside it. The CDC Field Epidemiology Manual (published 8 August 2024; checked 5 September 2026) treats population, variables and collection as design decisions.

Which start event, end event and clock should time-to-fill use?

Pick one start event and one end event, and name both. The common pairs are:

• approved or open requisition to written offer acceptance;
• publication to acceptance;
• search start to acceptance;
• approval to recorded employment start. This last pair is time-to-start, not the same metric.

SHRM's Time-to-Hire/Time-to-Fill Calculation Spreadsheet (16 November 2023; checked 5 September 2026) uses requisition opening and offer acceptance as a boundary, but it is not universal. Use timestamped fields with stable IDs, not vague labels such as "candidate hired." For multi-seat requisitions choose first fill, last seat or one row per position; do not average unlike episodes. A changed brief needs a new episode or bridge.

Choose one labelled clock:

• calendar elapsed (end minus start);
• working days, with a named holiday calendar;
• active elapsed, after approved pauses.

Store UTC plus the original offset or IANA timezone. Date-only fields lose precision; mark them Limited or unknown rather than assigning midnight. Pauses need a reason, approver, start and end, policy version and overlap handling; count overlaps once. An unresolved pause keeps the duration unresolved unless the cutoff rule closes it. A canceled requisition has no fill event: exclude it and report cancellation separately. Preserve history when reopening, and never treat withdrawal as zero days.

How to choose the time-to-fill cohort, denominator and source

Choose the cohort first: say whether it groups requisitions by start date or end date, which roles, statuses, worker types and places it covers, and how long you wait for corrections. A completed row needs:

• valid start and end events;
• the same process version;
• no unresolved pause or duplicate identity;
• documented rules for internal moves, agencies, rehires and multi-seat roles;
• a passed maturity window.

Open work belongs in aging, with an as-of timestamp and a separate denominator. The mean or median denominator is eligible mature episodes with valid endpoints, not candidates, applications, interviews, offers or all ATS rows. Show excluded, open, canceled, duplicate, missing and immature counts.

Name the system of record and extract query, filter, run time and source version. ATS data may need reconciliation with offer or HRIS acceptance data. Preserve both values, apply a predeclared priority or mark unresolved; do not choose the shorter duration. A spreadsheet is a calculation surface, not authoritative history.

How to handle time-to-fill revisions and decisions

Use Ready, Limited and HOLD statuses. Missing endpoints are unknown, not zero. For late acceptance, backfilled start, timezone correction or changed status definition, retain old/new values, reason, source timestamp, approver and affected cohorts; rerun results and trace the prior publication. The ONS revisions guidance (last updated 9 August 2024; checked 5 September 2026) supports this record but supplies no recruiting window. Version changes to events, pauses, timezone, cohort, source priority or statistic; bridge old periods where possible, otherwise start a new series.

Compare only like-for-like unit, events, clock, timezone, pauses, maturity, source, process version and statistic. Set a local trigger, then record one decision:

• Continue as scoped: a mature, reconciled result.
• Limited: a small cohort.
• Revise or bridge: a definition change.
• HOLD: missing endpoints, unknown pauses or unreconciled IDs.
• STOP: a request for a benchmark, candidate-quality judgment or prediction.

Fictional example: a July time-to-fill cohort

A July time-to-fill cohort is measured here under one written definition; the example is fictional, not a benchmark or forecast. Version TTF-v2 measures a requisition episode from approval/open (t0) to offer acceptance (t1), calendar days minus approved pauses. A July 2026 start-date cohort is extracted 7 August after a seven-day correction window; a 20-day trigger was predeclared.

The mature denominator is n = 3; active mean is 17.0 days and median 18 days, while gross mean is 18.3. The pause change needs a version break. This signal merits review of cancellation/open causes and a refresh after maturity; it says nothing about candidate quality, response, hiring success or market speed.

Copyable time-to-fill metric brief

Version/owner; unit and episode rule; scope; start/end fields and timezone;
clock and pause/overlap policy; cancellation/reopen/multi-seat rule;
cohort basis; maturity/correction window; denominator and statistic;
source/query/extract/process version; missing/duplicate/revision handling;
threshold and stop condition; status; next refresh and revision log.

Where Talent Summoner fits in time-to-fill reporting

Talent Summoner sits outside time-to-fill reporting: it supports sourcing and CV ranking, and supplies no metric events. Talent Summoner is our product for candidate sourcing and candidate ranking (checked 30 September 2026). Your team keeps the time-to-fill definition, the ATS and offer data, event reconciliation, pause rules and employment decisions. Candidate sourcing supports an approved brief; candidate ranking organises supplied material for human review. Neither supplies metric events or predicts speed, fit, interest, availability or hiring outcomes. Keep records and decisions with the authorised team.

Sources checked

• SHRM calculation spreadsheet, 16 November 2023; checked 5 September 2026. Boundary reference.
• Ashby Recruiting Operations Benchmarks, 7 May 2026, and Pin Recruitment Operations, 29 May 2026; checked 5 September 2026. Vendor context; no figures or product claims.
• ONS revisions guidance, 9 August 2024; checked 5 September 2026.
• CDC collecting and analyzing, 8 August 2024; checked 5 September 2026.

Keep the metric on HOLD when events, denominator, maturity, timezone, pauses or ownership are unclear.