The Four Recruiting Metrics That Actually Predict a Hire
Time-to-hire and cost-per-hire tell you what already happened. Four leading indicators tell you whether a search is working while you can still do something about it.
Most recruiting dashboards report time-to-hire and cost-per-hire. Both are real numbers. Both are useless while a search is running, because you only learn them after the outcome is already decided.
A search that will fail looks identical to one that will succeed for the first three weeks, if those are the only numbers you have. By the time time-to-hire tells you something, the information has no value.
What you want are leading indicators — numbers that move early and point at the specific thing that's broken. There are four.
Time to First Qualified Candidate
Not time to first message. Time until the hiring manager has someone real to look at.
This is the best early signal, for a reason that has nothing to do with speed: it tests whether the role definition survives contact with the market. If nobody qualified appears in the first few days, the problem is almost never sourcing effort. It's that the profile you described doesn't exist at the price and location you described it.
Learning that on day three rather than in week six is worth a great deal. The fix is a conversation about which constraint gives — seniority, compensation, location, or the domain background. Every week that conversation is delayed is a week of sourcing against an impossible brief. For a well-defined role, 48 to 72 hours is reasonable. Past a week, stop sourcing and revisit the brief.
Shortlist Acceptance Rate
Of the profiles you put in front of the hiring manager, what percentage do they agree to speak with.
This measures alignment between what the search is producing and what the manager actually wants — which is not the same as what they wrote down. It's the closest thing recruiting has to a quality metric with ground truth.
Below 30%, the scorecard doesn't encode the manager's real bar; sit down with three rejected profiles and ask what specifically was wrong, because those answers are missing criteria. Between 30 and 60%, workable but leaky, usually one criterion off. Above 80%, the definition is captured, and the manager's problem is now choosing rather than filtering.
The common mistake is treating a low acceptance rate as a candidate supply problem. It almost never is. It's a specification problem wearing a supply problem's clothes.
Positive Reply Rate on First Outreach
Not open rate, not connection acceptance. The percentage of contacted candidates who reply with genuine interest.
This measures two things at once, and the pattern separates them. High connection acceptance with low positive replies means the targeting is right and the message is wrong — the role isn't being made compelling, or it reads as generic. Low connection acceptance means the targeting is off, or the sender's profile doesn't carry credibility with this audience.
That second point deserves attention. Outreach from a hiring manager's own profile consistently outperforms outreach from a recruiter or a company account, often substantially. Same message, different sender, materially different result. Candidates respond to the person they'd actually work with.
Treat published benchmarks with suspicion. They vary enormously by market and seniority, and most are vendor marketing. Compare against your own baseline for comparable roles.
Where the Funnel Narrows
Individual conversion rates matter less than the shape of the funnel, because the shape gives you the diagnosis directly.
Narrow at sourcing means criteria too tight, or a thinner market than assumed. Narrow at outreach reply means targeting or messaging. Narrow at manager review means scorecard misalignment. Narrow at interview means the screen is passing people it shouldn't — your criteria measure the wrong things. Narrow at offer means compensation, or a role that sounded different from what it is.
Each of these has a completely different fix, and they get confused constantly. The most common error is responding to a narrow point late in the funnel by adding volume at the top, which makes the numbers look busier without improving anything.
One search in isolation is a small sample and a lot of noise. Comparing searches is where the signal lives. If four roles are running and three have healthy funnels while one narrows sharply at review, that isn't a market condition — it's one hiring manager's scorecard.
Two numbers not to optimise. Messages sent, which measures activity and degrades everything else when pushed. And average match score, which is dominated by the candidates you'll never contact; the shape of the top tail is what matters.
Umamy tracks these across every search you're running, in one view, so you can tell the difference between a market problem and a definition problem.