Cost per Applicant (CPA)

Anything you paid to generate applications in the measured window: job board slots and pay-per-click campaigns, programmatic advertising, sponsored social posts, careers-site advertising, event sponsorship where it drove applications. Agency fees do not belong here, because they buy submissions rather than applicants. Employer brand spend is the argued case. It clearly generates applications, but on a lag long enough that assigning this month's brand budget to this month's applicants is misleading. Most teams keep brand spend in a separate line and note that CPA therefore understates true attraction cost. Whatever you decide, write the rule down once and apply it every period. A CPA that changed because someone reclassified a budget line is worse than no CPA.

Because the cheapest applicants are, on average, the least qualified, and a metric that rewards cheapness will find them. Broadening targeting, loosening keywords, or advertising a specialist role on a general board all push CPA down while filling the queue with people who cannot do the job. Someone then has to review every one of them, and that review time is a real cost that CPA does not see. The honest way to use the metric is as a pair: cost per applicant beside applicants per hire, or beside cost per hire. If CPA fell and applicants per hire rose by more, the campaign got worse and the dashboard says it got better. Judge attraction spend on qualified applicants where you can define qualified.

Duplicates first. The same person applying to four openings is four applications and one applicant, and the two are constantly confused in reporting. Decide which you are counting and label the metric accordingly. Bot and scraper traffic inflates counts on public forms, particularly on high-visibility postings, and it arrives in bursts that look like a successful campaign for about a day. Incomplete applications are the third category: someone who started a form and abandoned it halfway is not an applicant, though some systems count them. Deduplicate on email and phone, exclude submissions that never completed, and look at the count by hour when something spikes. A clean denominator changes CPA more than most media optimisation does.

Use it to move money between sources, not to shrink the budget by default. Compare CPA per source against that source's downstream yield, and the picture usually splits into three groups: expensive sources that deliver hires, cheap sources that deliver noise, and a middle band worth testing further. Shift spend toward the first group even where it raises headline CPA, because the number that pays the bills is cost per hire. For high-volume roles, small changes to the posting and the application form often beat any media buy, since a shorter form lifts completed applications at no additional media cost. Multi-board job posting software makes per-source spend and applicant counts comparable, which is the precondition for any of this.