Recruiting analytics is mostly a small set of measures that are easy to collect and rarely acted on. The useful ones connect the process to its outcome rather than describing its activity, and at low hiring volume most of them need reading as trends over a year rather than as monthly figures.
The measures that connect to an outcome
Offer acceptance rate, which tells you whether your offers are competitive. Early leaver rate by source, which tells you which channels produce people who stay. And time from application to first response, which is the single measure most correlated with losing good candidates. None requires a specialist tool.
The measures that mislead
Cost per hire, which moves with your channel mix and vacancy profile far more than with anything you control. Number of applications, which rewards advertising widely rather than hiring well. And time to fill taken as a monthly figure at low volume, where two slow hires make a quarter look broken.
Monitoring for fairness
Progression rates by group at each stage of the process, reviewed regularly. This is the analysis that finds a problem while it is still cheap to fix, and it is straightforward to produce from an applicant tracking system. Not looking is a choice, and the discovery otherwise happens in a complaint.
Questions people ask about recruiting analytics
How much hiring volume is needed?
For fairness monitoring, aggregate a year or more at small volumes. For response time, any volume, because it is measured per candidate.
Is cost per hire worth tracking?
As a budget figure yes, as a performance measure no. It moves with factors outside the recruiter's control.
Where does the data come from?
The applicant tracking system for process measures and the HR system for early leavers. Joining the two is where the useful analysis starts.