Skip to content
Hiring teams7 min read

How to Measure Quality of Hire

A practical quality of hire metric: four components, when to define them, and how to check whether your scorecard predicted who did well.

By the HireRabbit.AI team · Published · Last updated

A recruiter links a stack of scored candidate cards by a thread to a growth chart of new hires a year later
On this page
  1. Why quality of hire is hard to measure
  2. What the research and surveys say
  3. A quality of hire composite you can run
  4. Define it before you open the role
  5. Close the loop to your screening and interview scores
  6. A worked example: six support engineer hires
  7. Data quality caveats
  8. How this maps to HireRabbit.AI

Most hiring teams can quote their time to fill and cost per hire. Ask whether last year's hires turned out well and the answer is usually a story about two or three people. Quality of hire is the metric everyone agrees matters most and few teams track.

This guide covers why quality of hire is hard to measure, what the research says, and a four-part composite you can run in a spreadsheet. It then shows how to check whether your screening and interview scores predicted who did well, with a worked example of six support engineer hires.

Why quality of hire is hard to measure

Speed and cost metrics close when the offer is signed. Quality of hire only starts then, which creates two problems.

The lag problem

The useful signals arrive months after the decision. A 90-day manager check-in is the earliest point anyone can say much, and first-year retention takes a full year. By then the recruiter has run dozens of other searches.

Shanil Kaderali of recruitment outsourcing firm Pierpoint International made the point to SHRM in 2015: the first analysis "means little, as there's nothing to compare." Plan for a year or more of data before expecting firm answers.

The attribution problem

A new hire's results depend on more than the hiring decision. The manager, onboarding and the team all play a part. Lars Schmidt, founder of Amplify Talent, put it plainly in the same SHRM article: "You can have an amazing hire and a horrible manager."

So a quality of hire score is evidence about the hire and their first months together. Read it in aggregate, across many hires, and look for patterns rather than verdicts on single people.

What the research and surveys say

The measurement gap is well documented. SHRM's 2025 benchmarking survey of 2,371 members found that only 20% of organizations track quality of hire.

LinkedIn's Future of Recruiting 2025 report found that 89% of talent acquisition professionals agree measuring quality of hire will become more important, but only 25% feel highly confident their organization can do it well. The measures they use most:

  • Job performance ratings, used by 66%.
  • New hire retention, used by 60%.
  • Hiring manager satisfaction, used by 44%.

LinkedIn's advice is that there is no one formula, and that "the most important thing is to measure them consistently." The report also describes Uber's approach: define success profiles from top performers, assess candidates against them, then validate hires with post-hire surveys of hiring managers.

Selection research shows why that last step matters. Sackett, Zhang, Berry and Lievens' 2022 re-analysis in the Journal of Applied Psychology found that the validity of most selection methods had been overestimated, with revised estimates .10 to .20 lower. Structured interviews still ranked first at .42, ahead of job knowledge tests at .40 and cognitive ability tests at .31. Your own process may do better or worse than that average, and the only way to know is to compare pre-hire scores with outcomes.

A quality of hire composite you can run

No single measure works alone. Manager satisfaction drifts toward whoever is easy to manage, retention ignores how well people do the job, and ratings arrive late. A composite of four parts balances them.

ComponentWhenScore (0 to 100)Weight
Hiring manager ratingDay 90Anchored 1 to 5 rating: 1 = 0, 2 = 25, 3 = 50, 4 = 75, 5 = 10030%
Ramp timeWhen the hire first meets the agreed targetOn target = 100, up to 4 weeks late = 50, later = 020%
First-year retentionMonth 12Still employed = 100, left = 020%
Performance ratingFirst formal reviewSame mapping as the manager rating30%

Quality of hire = 0.3 × manager rating + 0.2 × ramp + 0.2 × retention + 0.3 × performance. Report the 90-day parts early as an interim view, and the full composite once the year is up.

Treat the weights as a starting point. A February 2026 SHRM preview of a Talent 2026 session quoted Shiran Danoch of Informed Decisions recommending performance ratings at six and 12 months, time to productivity, retention and manager satisfaction, customized to the role. A sales team might swap ramp time for quota attainment.

Define it before you open the role

The definition has to exist before anyone sees a candidate. Otherwise "a good hire" quietly becomes "the person we picked." In the same SHRM preview, Danoch argued that hiring managers, talent acquisition and leadership should "agree on these metrics before the first candidate interview."

Add four lines to the intake you already run for the job description:

  1. The 90-day question. One question the manager answers at day 90, with anchors at 1, 3 and 5. For example: "How independently does this person handle a standard escalation?"
  2. The ramp target. One observable milestone and the expected week, such as "closes tier 2 tickets without review by week 10."
  3. What counts as a leaver. Decide whether a relocation or an internal transfer counts as a loss.
  4. The weights. Agree them, date them, and leave them alone for the year.

Store this with the role's scoring sheet. If it changes, keep the old version, as you would for any hiring decision record.

Close the loop to your screening and interview scores

This is where quality of hire earns its keep. Once a cohort has its scores, line up each hire's pre-hire scores beside the composite and ask: did the people we scored highest turn out best? Run three comparisons:

  • Resume screen score. If it does not track quality of hire, look at what your resume scoring rubric is rewarding.
  • Interview average. If this is flat, your structured interview scorecard anchors may be too vague, or one interviewer may rate everyone the same.
  • Each competency. Often one carries most of the signal and another carries none. The one that carries none is a candidate for rewriting or dropping.

For a handful of hires, sorting by score is enough. With 20 or more in the same role, a spreadsheet function such as CORREL gives one figure to track from year to year.

A worked example: six support engineer hires

A 200-person software company hired six customer support engineers last year. Each had a 0 to 100 resume screen score and a 1 to 5 interview average from a five-competency scorecard. The team agreed the composite above at intake, with a ramp target of 10 weeks.

HireScreen scoreInterview avgManager, day 90Ramp (weeks)Here at 12 monthsPerformanceQuality of hire
A864.248Yes485
B813.4312Yes360
C744.457Yes5100
D903.0316No330
E683.8410Yes485
F773.6413Yes475

Check the formula on hire A. A manager rating of 4 maps to 75, ramp at 8 weeks beats the target for 100, retention is 100 and performance of 4 maps to 75. That gives 0.3 × 75 + 0.2 × 100 + 0.2 × 100 + 0.3 × 75 = 85.

Now sort. By interview average the order is C, A, E, F, B, D, and quality of hire falls in the same order: 100, 85, 85, 75, 60, 30. By screen score the order is D, A, B, F, C, E, and the top screened hire, D, has the lowest quality of hire.

The team checks what D's resume scored on: the most years in support and a large-company background, both of which the rubric rewarded. The interview, which asked for escalations handled end to end, caught what the screen missed. The team moves weight from years of experience to evidence of owning escalations, dates the change and checks again with the next cohort.

Data quality caveats

A composite is only as good as its inputs. Keep these limits in mind before anyone acts on the numbers.

  • Ratings say a lot about the rater. Scullen, Mount and Goff's 2000 study in the Journal of Applied Psychology found that idiosyncratic rater tendencies accounted for 62% and 53% of the variance in performance ratings across two large data sets, against 21% and 25% for actual performance. Anchored questions help but do not remove this.
  • You only see the people you hired. Rejected candidates never get a score, so the comparison runs on a narrower range than the applicant pool. This range restriction is the issue at the center of Sackett and colleagues' re-analysis, so read a weak result with care before scrapping a step.
  • Small numbers mislead. Six hires can suggest where to look but prove nothing. Pool similar roles or several years before changing a process.
  • Retention mixes causes. A partner's relocation and a poor first review both score zero. Record the reason for each departure and apply exclusions consistently.
  • Check outcomes by group. If scores differ by gender or ethnicity, the cause may be the ratings rather than the hires. Use the checks in the adverse impact analysis guide.
  • Watch for drift. Once managers know their 90-day rating feeds a recruiting metric, ratings can creep up. Review the distribution each quarter.

How this maps to HireRabbit.AI

Closing the loop needs a pre-hire score that means the same thing for every candidate. In HireRabbit.AI, every resume gets four area scores for skills, experience, projects and education, and the overall score is computed from those four using weights that stay the same for every applicant. That gives each hire a stable screen score to set beside their quality of hire a year later.

Any move of a single candidate carries a written reason, so when a comparison turns up a surprise like hire D, you can see what the team was thinking at the time. The how scoring works page explains the area scores and weights.

Put this into practice.

Free during invite-only early access. All you need is a work email, and a person reviews every request.