Recruitment efficiency benchmarks: how to compare hiring performance
On this page 14 sections
There is no universal good time to hire, cost per hire, or applications-per-hire ratio. A useful recruiting benchmark compares the same metric definition, role family, seniority, location, employment type, and time period. It also pairs speed and cost with hiring quality, candidate outcomes, and fairness.
That answer is less convenient than a single industry average, but it is safer and more useful. A company can reduce calendar days by cancelling difficult searches, lowering assessment standards, or shifting work to agencies. None of those changes necessarily improves recruitment.
This guide explains what current public evidence can establish, what vendor datasets can and cannot establish, and how to build an internal benchmark that leads to a specific operating decision.
Benchmark snapshot and evidence boundary
The strongest broad US reference in this review is the SHRM 2025 Recruiting Benchmarking Report. SHRM surveyed 2,371 members between January 9 and March 3, 2025. Results were not weighted, and the number of responses varies by metric. That makes the report a useful reference cohort, not a universal target.
SHRM reports average time to fill of roughly one and a half months for both executive and nonexecutive roles. It reports average cost per hire of $5,475 for nonexecutive roles and $35,879 for executive roles. Only 20% of participating organizations said they tracked quality of hire.
Those figures establish two useful points. First, recruitment takes material time and money. Second, a comparison is incomplete unless the employer knows whether its own numerator, denominator, role mix, and clock match the survey definition.
Operational reports from recruiting-software companies can add detail because they observe workflow events. They are still vendor cohorts. Their customers, configuration, data quality, and product usage may not represent the wider labor market.
Define each clock before comparing it
Teams often use the same label for different intervals. A change in definition can look like a performance improvement even when the process has not changed.
| Metric | Recommended start | Recommended end | Decision it supports |
|---|---|---|---|
| Requisition approval time | Request submitted | Requisition approved | Whether workforce planning or approvals delay hiring |
| Time to fill | Requisition approved | Offer accepted | How long a business need remains open |
| Time to hire | Candidate enters active consideration | Offer accepted | How quickly the process converts a known candidate |
| Stage time | Entry into a named stage | Exit from that stage | Where candidates and reviewers wait |
| Time to start | Offer accepted | First working day | Whether notice periods, checks, or onboarding create delay |
| Hiring-manager decision time | Final assessment completed | Decision recorded | Whether decision ownership is clear |
Record the definition in the dashboard itself. Also state whether weekends, evergreen requisitions, internal transfers, cancelled roles, reopened roles, and multiple hires on one requisition are included. Without those rules, a month-over-month comparison can be misleading.
Read the SHRM figures as reference points
The SHRM report is most useful as a sense check on magnitude and measurement maturity. It does not provide a role-by-role service-level agreement for every employer.
An employer whose nonexecutive cost per hire exceeds the SHRM average has not automatically found waste. It may hire scarce specialists, operate across several countries, pay relocation costs, use external search firms, or include more internal labor in its calculation. An employer below the average may omit interview time, technology allocations, referral payments, or failed-search costs.
The quality-of-hire result is more actionable. If only one in five respondents tracks the measure, a team that optimizes speed without an outcome measure has a common but serious blind spot. The first improvement should be a stable quality definition, not a lower time target.
Treat vendor datasets as operational cohorts
Gem’s 2025 recruiting benchmarks describe a vendor dataset covering 140 million applications, 14 million candidates, and 1.3 million hires. Ashby’s recruiting operations benchmarks describe 54 million applications and 93,000 jobs from January 2021 through March 2026.
These are valuable disclosures because the sample size and data source are visible. They are not random samples of all employers. They mainly describe organizations using each vendor’s platform and the records those customers configured well enough to analyze.
Use such reports to generate questions:
- Is one stage taking unusually long for a comparable role family?
- Is the team advancing many applicants but hiring few?
- Do sourced and inbound candidates move through the funnel differently?
- Does a change appear across several cohorts or only in one vendor’s population?
Do not copy a vendor percentile into a company-wide target without reconstructing its metric definition and relevant cohort.
Use role-level cohorts
Industry is often too broad to explain recruiting difficulty. A hospital may hire nurses, data engineers, facilities staff, finance leaders, and executives. A software company may run high-volume support hiring alongside a small number of specialized security searches.
A practical benchmark hierarchy is:
- Role family and required credential.
- Seniority and decision scope.
- Location and labor-market reach.
- Employment type and work arrangement.
- Hiring channel, such as internal, referral, sourced, inbound, agency, or campus.
- Business unit and hiring manager.
- Industry, company size, and external reference cohort.
Start with internal historical cohorts that meet these conditions. Add external data only when its definition and population are sufficiently comparable. If a cohort is small, show the count and use a rolling period rather than presenting an unstable percentage as a trend.
Separate labor-market conditions from process performance
The US Bureau of Labor Statistics Job Openings and Labor Turnover Survey definitions cover job openings, hires, quits, layoffs, and other separations at an economy or industry level. JOLTS is useful context for labor demand and movement. It does not reveal whether a particular employer’s interview scheduling or approval process is efficient.
Use macro data as an explanatory layer, not a substitute for workflow evidence. A lower hiring rate could reflect weaker demand, fewer approved requisitions, a scarce talent pool, a slower internal process, or some combination. Pair macro indicators with the company’s openings, applications, stage times, offers, starts, and cancellation reasons.
Calculate cost per hire consistently
A defensible internal cost-per-hire calculation begins with a written cost boundary:
cost per hire = (included internal recruiting costs + included external recruiting costs) / completed hires
Internal costs may include recruiting compensation, allocated technology, employee-referral payments, interview labor, and recruiting operations. External costs may include agencies, advertising, assessments, background checks, events, relocation, or immigration services. Decide whether to include each item, document the choice, and keep it stable across periods.
Vacancy cost requires a separate model. Do not apply a generic dollar-per-day estimate to every role. Estimate only costs that the business can support, such as overtime, contractor coverage, delayed revenue, service-level penalties, or capacity lost while the role is open. Show assumptions and a range. Avoid counting the same effect in both vacancy cost and lost revenue.
Cost per hire is diagnostic when paired with channel and outcome. A higher-cost source that produces durable, high-performing hires may be more efficient than a low-cost source with poor retention or low offer acceptance.
Make quality of hire measurable
Quality of hire should be a cohort outcome, not a recruiter’s subjective score. Choose a small set of measures that are available for most hires and have a plausible link to job success.
Possible components include:
- completion of a structured probation or ramp milestone;
- job-relevant performance at a defined review point;
- regrettable attrition within a specified period;
- hiring-manager assessment against criteria set before the search;
- new-hire assessment of role clarity and process accuracy.
Publish the formula and missing-data rule. Compare like roles and start dates. Do not evaluate recent hires on a milestone they have not had time to reach. Keep performance and retention data access-controlled, and do not infer quality from protected characteristics or proxies.
A useful review shows speed, cost, and quality side by side. If time to fill falls while early attrition rises, the intervention needs investigation rather than celebration.
Measure funnel conversion without rewarding rejection
For every conversion rate, name the start and end states. Examples include qualified applicant to first interview, first interview to final assessment, final assessment to offer, offer to acceptance, and acceptance to start.
Raw applications per hire is especially easy to misuse. More applications can increase workload without increasing qualified supply. A low ratio can mean precise sourcing, a narrow top-of-funnel definition, or missing records. A high ratio can mean broad reach, an easy application flow, duplicate submissions, or poor targeting.
Track both counts and rates. Add median stage time and a reason taxonomy for rejection, withdrawal, offer decline, cancellation, and no-show. This makes a funnel actionable. For example, a stable qualified-applicant rate with a worsening interview-to-offer rate points to assessment or decision design, not necessarily sourcing.
Bound early-career comparisons carefully
The NACE 2025 Recruiting Benchmarks executive summary covers college recruiting. It reports 197 member respondents and 15 nonmember respondents, with a 24.9% response rate. Its measures are useful for campus and early-career programs, not for executive search or all corporate hiring.
NACE’s summary reports that candidates received about nine days on average to decide on an offer. That can inform an early-career candidate communication policy, but it is not a universal deadline. Employers should consider local law, institutional guidance, candidate circumstances, and the time needed to understand compensation and work conditions.
Treat fairness and accessibility as performance constraints
Faster screening is not efficient if it creates unlawful or job-irrelevant exclusion. The US Equal Employment Opportunity Commission states that employment practices can be unlawful when they discriminate because of protected characteristics, including practices that appear neutral but have a disproportionate negative effect and are not job-related and necessary. The agency’s prohibited employment policies and practices guidance applies across recruitment, testing, and hiring.
Monitor selection rates and process access by stage where lawful and appropriate. Investigate meaningful differences with qualified legal and industrial-organizational expertise. A simple threshold is a screening signal, not proof that a process is fair or unlawful.
Also measure accommodation requests, completion failures, appeal or review requests, and manual overrides for automated tools. These are operating signals. They can reveal that an assessment, scheduling flow, or communication channel is preventing qualified people from completing the process.
Use a diagnostic matrix before changing targets
| Observed pattern | Evidence to inspect | Plausible intervention | Guardrail |
|---|---|---|---|
| Approval time is rising | Requests by business unit and approver | Clarify workforce-plan authority and escalation | Do not approve roles without budget or job scope |
| Review time is rising | Queue age, reviewer load, qualified rate | Set ownership and review cadence | Audit whether filters exclude qualified applicants |
| Interviews per hire are rising | Interview count, score agreement, duplicate topics | Remove redundant rounds and structure questions | Preserve job-related evidence and accommodations |
| Final decision time is rising | Time from completed evidence to decision | Define decision owner and decision meeting | Do not force decisions before required evidence exists |
| Offer acceptance is falling | Decline reasons by role and location | Improve compensation calibration and role clarity | Do not pressure candidates or hide material terms |
| Time falls but early attrition rises | Start cohorts and quality outcomes | Review assessment validity and expectation setting | Do not optimize speed in isolation |
The matrix connects a metric to evidence and a bounded change. It also prevents an average from becoming a blanket demand to work faster.
Build a monthly recruiting scorecard
Use medians and percentiles for duration metrics because a few long-running searches can distort an average. Keep counts beside percentages and report small cohorts cautiously.
A compact monthly scorecard can include:
- approved requisitions, hires, starts, cancellations, and open roles;
- median and 75th-percentile time to fill by role family;
- median stage time and queue age;
- qualified-applicant, interview, offer, acceptance, and start conversion;
- cost per hire by channel using a stable cost boundary;
- quality outcomes for mature cohorts;
- candidate withdrawal and offer-decline reasons;
- selection-rate and accommodation indicators reviewed under the applicable legal framework;
- data completeness, missing outcomes, and definition changes.
Annotate policy, tooling, labor-market, and hiring-plan changes. A trend line without those events can assign credit or blame to the wrong cause.
A 30-day benchmark reset
Week 1: define. Agree on the business question, metric dictionary, inclusion rules, cohort dimensions, and data owners.
Week 2: reconcile. Sample records from the applicant tracking system, finance, HR information system, and offer process. Resolve missing timestamps, duplicate candidates, reopened requisitions, and inconsistent stage names.
Week 3: segment. Produce internal baselines by role family, seniority, location, and source. Add an external reference only where its population and definition are visible.
Week 4: intervene. Select one bottleneck, name an owner, define the expected mechanism, choose a quality and fairness guardrail, and set a review date. Avoid changing several stages at once if the team needs to learn what caused the result.
Frequently asked questions
What is a good time to hire?
It is the shortest interval that consistently produces qualified, accepted, and durable hires for a comparable role cohort without compromising fairness or candidate access. An external average is a reference point, not the answer.
Should we use time to fill or time to hire?
Use both. Time to fill measures how long the business need remains open. Time to hire isolates the candidate process after a person enters active consideration. Stage times explain which part changed.
Is a lower cost per hire always better?
No. Cost can fall because the employer shifts work outside the calculation, stops investing in scarce roles, or hires people who leave quickly. Review cost with channel, quality, retention, and capacity.
Can vendor benchmark data be trusted?
It can be useful when the vendor discloses the sample, period, definitions, and exclusions. Label it as a vendor cohort and test whether its customer population resembles the decision at hand.
Bottom line
Recruitment benchmarking is a measurement discipline, not a league table. Define the clock, segment the work, disclose the cohort, and pair speed and cost with outcomes and fairness. The best benchmark is the one that identifies a real constraint and supports a controlled operating change.