Time-to-hire measures how fast your process runs; quality-of-hire measures whether the decisions were right. Speed metrics dominate hiring dashboards because they are easy to collect and available immediately, while quality signals require waiting months — which is exactly why most teams optimise the wrong thing.
- Time-to-hire and time-to-fill are different metrics. Most teams report one and mean the other.
- Every speed metric needs a quality metric paired with it, or it will be gamed — usually unintentionally.
- Quality-of-hire needs a definition agreed before you hire, not an argument afterwards.
- Assessment-score-to-performance correlation is the highest-value number in hiring and almost nobody has it.
- Track completion rates by stage. Candidates leaving your process is data about the process.
Why the wrong metric wins
Not because anyone believes speed matters more than quality. Because of latency.
Time-to-hire is available the moment someone accepts. Quality-of-hire is available in six to twelve months, and only if you recorded the right things at the start. Build a dashboard from what is available today and you get a speed dashboard — then you manage what is on the dashboard.
This is worth naming plainly because it dissolves the usual argument. Teams are not choosing speed over quality; they are choosing measurable now over measurable later, and then inheriting the incentives.
The fix is unglamorous: start recording the slow signals now, so that in a year the quality metrics exist and can compete for attention.
Get the speed metrics right first
They are worth measuring — they are just diagnostics of throughput, not of judgement. And most teams conflate two different ones:
| Metric | From | To | Diagnoses |
|---|---|---|---|
| Time-to-fill | Role approved | Offer accepted | The whole system, including sourcing and requisition delay |
| Time-to-hire | Candidate enters process | Offer accepted | How fast you move a person once you have them |
| Stage time | Enters stage | Leaves stage | Where the queue actually is |
| Time-to-first-response | Application | First human contact | Candidate experience, and drop-off risk |
Stage time is the useful one. An aggregate of 42 days tells you nothing actionable. Stage time tells you that candidates sit eleven days waiting for interview scheduling, which is a calendar problem with a calendar fix.
Time-to-first-response is the underrated one. It predicts drop-off, and it is the metric candidates actually experience. It is also the leading indicator for ghosting — if first response is slow, final response is usually absent.
Defining quality-of-hire
The reason quality-of-hire rarely gets measured is that it needs a definition, and the definition needs agreeing before you hire. Afterwards it becomes an argument about whether a specific person worked out.
Pick two or three of these, define them precisely, and hold them stable for at least a year:
- Performance rating at 6 and 12 months. Imperfect and biased by the same forces as hiring, but consistent and already collected.
- Hiring manager would hire again. One question, asked at 6 months, answered yes or no. Crude and surprisingly informative.
- Ramp time to full productivity. Requires a definition of "full productivity" per role. Worth the effort; it is the metric that most directly captures whether the assessment was right.
- First-year retention, split by voluntary and involuntary. Critically, split. A voluntary exit at eight months and a performance termination at eight months mean opposite things about your hiring, and an aggregate retention number hides both.
- Assessment-score-to-performance correlation. The most valuable and the rarest. Discussed below.
Two warnings. Do not build a composite "quality of hire index" — a weighted blend of five things nobody can interpret is worse than two numbers you understand. And be honest about attribution: a hire failing in a role with no onboarding, an absent manager and shifting goals is not necessarily a hiring failure. Poor quality-of-hire numbers are sometimes a management finding.
The metric almost nobody has
Does your assessment score predict subsequent performance?
If the answer is no, every other metric in your funnel is measuring the throughput of a process that does not distinguish good candidates from bad ones. Faster, cheaper, more diverse at the top of a funnel that cannot tell who is capable produces the same hires, sooner.
Nobody has this number for a mundane reason: it needs assessment scores and performance outcomes joined by person, and those live in different systems owned by different teams, and nobody joined them at the time.
Start now. A single table — candidate identifier, role, assessment score by dimension, hire date, performance indicator at six and twelve months — is enough. In eighteen months it is the most valuable dataset in your hiring operation, and it is the only way to answer the validity question about your own process.
Pair every metric with its counterweight
Any single metric optimised in isolation degrades. Not through cynicism — through ordinary local optimisation by people trying to hit a target.
| Metric | What it becomes alone | Pair it with |
|---|---|---|
| Time-to-hire | Fewer stages, less assessment, rushed decisions | Quality-of-hire, 12-month retention |
| Cost-per-hire | Cheap sourcing, no assessment investment | Quality-of-hire, score-to-performance |
| Offer acceptance rate | Only offer to certain-yes candidates | Pipeline diversity, selection rates by group |
| Application volume | Vanity traffic, unread applications | Qualified-application rate, completion rate |
| Assessment pass rate | Threshold lowered until it clears | Score distribution, adverse-impact ratios |
| Diversity of hires | Focus at the top of the funnel only | Stage-level selection rates, retention by group |
The pairs are the actual system. A single number on a dashboard is an invitation.
Two metrics almost everyone is missing
Completion rate by stage, split by group. Candidates abandoning a stage have been filtered — just not by a decision you recorded. A six-hour take-home with a 40% completion rate has selected on available free time, and that never appears in a selection-rate analysis. This is also the blind spot in most adverse-impact analysis.
Override rate. How often does a human decision diverge from the tool's recommendation? A rate near zero means your reviewers are ratifying rather than deciding, which matters for decision quality and for your human-oversight obligations.
A dashboard worth having
Throughput — stage time, time-to-first-response, time-to-fill. Fairness — selection rate and completion rate by group, per stage. Judgement — score distribution, reviewer agreement, override rate. Outcome — would-hire-again at 6 months, ramp time, 12-month retention split by exit type. The keystone — assessment score versus performance.
Five groups, roughly a dozen numbers, and only the last group takes a year to populate — which is why you start it today.
What CalHire gives you by default
Because assessment, decisions and outcomes live in one auditable flow, several of these stop being data projects:
- Hiring analytics from Growth upward, with stage-level reporting rather than aggregates only.
- Explicit, visible score weights per role — the default composite is test 40 / interview 35 / role-fit 25 — so a score distribution is interpretable rather than opaque.
- Every person-affecting decision logged and attributed to the human who made it, in a tamper-evident audit trail. That is what makes an override rate computable at all.
- Bias-audit exports and stage-level selection reporting from the compliance console, produced from the same records as the decisions.
- Data export and API for joining assessment data to your HRIS performance data — the keystone metric, which no hiring platform can compute for you because the performance half lives in your systems.
See /features for the analytics and audit surfaces, and /pricing for which tier includes export and API access.
If you only change one thing: pick one quality metric, define it today, and start the score-to-performance table. Speed metrics will still be there in a year. The quality data only exists if you started.
Frequently asked questions
- What is the difference between time-to-hire and time-to-fill?
- Time-to-hire usually measures from a candidate entering your process to accepting an offer — it is a measure of how fast you move a person through. Time-to-fill measures from the role being approved to someone accepting — it includes sourcing time and any delay in getting the requisition opened. They diagnose different problems, and reporting one while meaning the other is a common source of confused conclusions.
- How do you measure quality of hire?
- Pick two or three indicators, define them before you hire, and hold them stable: performance rating at six and twelve months, whether the hiring manager would hire the person again, ramp time to full productivity, and retention through the first year distinguishing voluntary from involuntary exits. None is perfect; together they are far better than a debate about who was a good hire.
- Why do teams over-focus on speed?
- Because speed data arrives immediately and quality data arrives in six to twelve months. A dashboard populated with what is available will be a speed dashboard. That is a measurement-latency problem, not a values problem — and the fix is to start recording the slow signals now so they exist later.
- What is the single most valuable hiring metric?
- The relationship between assessment scores and subsequent job performance. It tells you whether your evaluation is measuring anything real, which determines whether every other number is meaningful. Almost no organisation has it, because it requires linking two datasets nobody joined at the time.