Cost per application became popular for a good reason. It is easy to calculate, easy to compare and available quickly enough to change a campaign while it is still running.
The problem starts when CPA gets promoted from an acquisition metric to a business verdict.
A cheap event can create expensive work.
Suppose one source delivers applications for $9 and another for $28. The $9 source looks like the obvious winner until the team learns that only 6% of those applicants reach recruiter review, compared with 35% from the second source.
Now add recruiter time, assessment cost, duplicate cleanup and hiring-manager effort. The ranking can reverse completely.
Keep CPA. Put it in its place.
CPA tells you what you paid for the application event. It does not tell you whether the application was usable, whether the candidate was reachable, whether the recruiter reviewed the record, whether an interview happened or whether the source created incremental value.
Pair it with:
- Cost per minimum-qualified candidate
- Cost per recruiter-reviewed candidate
- Cost per interview
- Cost per accepted offer
- Processing hours per hire
- Time to first usable slate
The farther down the funnel you can reliably measure, the more meaningful the comparison becomes.
Do not punish sources for things they cannot control.
Source quality can appear weak because hiring managers delay review, ATS source fields get overwritten or candidates encounter a broken application experience. Recruitment marketing ROI is a connected measurement problem.
That is why the answer is not simply “optimize to hires.” The employer first has to know whether the data connecting source to hire is trustworthy enough to carry that decision.
Attribution can make a good source disappear.
A candidate may see a display ad, read an employer article, search the company later, return through an organic listing and apply directly. The ATS may credit the final visit because it stores one source. That does not prove the earlier exposure was irrelevant.
Use attribution as evidence, not certainty. Where major budget decisions depend on it, supplement source fields with campaign parameters, assisted-touch data, market comparisons or controlled tests.
Recruiter labor belongs in the economics.
Recruiter time is often ignored because it is already inside payroll. That does not make it free. If one source requires hundreds of additional review hours, the employer is paying for that difference somewhere.
Use reasonable estimates. You do not need to track every mouse click. Sample review time, exception handling and downstream processing. Even a directional labor estimate can reveal that the $9 application is not really a $9 acquisition.
Use a decision threshold before you move the money.
Before changing budget, agree on the evidence required. If one source has a lower CPA but materially worse interview conversion for three consecutive cohorts, that may justify a change. If only application volume differs and downstream data is missing, the correct decision may be to repair measurement first.
What should change this week?
Choose your three largest sources. Put CPA beside cost per recruiter-reviewed candidate and cost per interview. If you cannot calculate either downstream number, write that gap on the report. The missing evidence is part of the decision.
Next step: Use The New Hiring Math calculator and read Recruitment Marketing ROI Is Not a Media Metric.