Every number on this site, with its receipts
The rules behind every number on this site: what we count, what we refuse to count, and where each published figure comes from. Candidate NPS and interview completion rate in full.
Hiring software is sold on metrics that are almost never defined. A "93% candidate satisfaction" figure with no question wording, no sample and no response rate is not evidence of anything — and it is the industry norm. These pages exist so that any number we publish can be checked, including by someone trying to disprove it.
The rules
- A number is published with its denominator or not at all. An NPS with no response count and a completion rate with no sample size are decoration. Every figure on this site carries the population it was computed over, including when that population is small enough to be unflattering.
- Measured and claimed are different words. Where we have not instrumented something yet, the marketing pages make a design commitment and say so — they do not quote a number we have not measured. Our own /vs hub labels vendor-reported metrics "vendor-reported, unaudited", and that rule applies to us first.
- The definition lives next to the code that computes it. Each methodology page mirrors the module that produces the figure, and the two are changed together. A definition maintained separately from its implementation is a definition that is already wrong.
- We state what we will not publish. Per-recruiter and per-campaign breakdowns of candidate sentiment are not published, in aggregate or otherwise. A methodology that only describes what it reports is incomplete.
What we measure
Candidate NPS · interview completion rate
Candidate experience methodology
The exact survey question and scale, who counts as part of the sample, the response rate, and the two things we will not do with the numbers. The live figures on the page are rendered by the same aggregation that feeds the public metrics endpoint, so the page cannot drift from what the API reports.
Looking for what we commit to rather than how we measure it? That is our candidate experience standards. For the term itself, see the candidate experience definition.
Found something wrong
If a figure here does not reconcile, we would rather hear it than not. Write to support@vettika.com and the correction goes on the page, dated.
See the output before you trust the method
A public sample report: the scored rubric and full transcript a 10-minute AI first round hands back.