Evidence you can inspect. Recommendations you can challenge.
How we review recruitment advertising, evaluate published studies, and develop campaign experiments without confusing observable activity with actual hiring performance.
What we count as evidence
Observed: A dated and source-linked advertisement, job description, public compensation disclosure, creative format, or application experience. We record what is visible, not inferred campaign performance.
Research-supported: A published study with a verifiable source, defined methodology, stated population and limitations.
Recommended test: A hypothesis with a proposed comparison, measurement plan and decision rule. A hypothesis is not a measured outcome.
Caliber-tested: A measured experiment conducted with permission and documented methodology. We do not use this label without actual campaign data.
How an employer ad review works
- Define the hiring roles, geography, observation window and inclusion rules before collecting ads.
- Identify advertisers and review supported public ad libraries and corresponding employer careers pages.
- Record advertiser identity, creative text and format, observation date, available run information, advertised role, and destination link.
- Code the primary message and any secondary themes, such as pay, scheduling, team support, growth or job specifics.
- Deduplicate creative variations where appropriate and document missing information and verification limits.
- Independently review a sample of coded observations and publish sample size, scope and limitations.
Ad visibility, duration and frequency are not sufficient evidence of qualified applications, hires, cost per application or return on investment. Those require permissioned campaign and recruiting-outcome data.
How recommendations become experiments
We separate a messaging hypothesis from a confirmed employment claim. Employers must verify compensation, schedules, benefits, staffing and advancement opportunities before using them in ad creative.
A proper test defines its comparison, eligible audience, geographic scope, budget, destination experience and primary KPI in advance. Where possible it holds other factors constant and uses qualified completed applications rather than clicks as the main outcome. Small samples and uncertain findings are reported as inconclusive.
What our research does not promise
Public advertiser libraries do not provide a complete view of the market, and most do not expose U.S. commercial recruitment campaign conversion performance. Source access varies by platform and permission. We will not fabricate competitor ads, market share, hiring results or campaign ROI to fill gaps.
Original experiments and competitive audits are clearly marked as proposed until evidence has been collected, reviewed and published. Platform access and employer data usage must respect applicable terms and permissions.
Data sources and citations
Our first healthcare messaging brief uses published experimental research and identifies which findings come from U.S. or non-U.S. samples. Employers and researchers can follow the underlying source links from the individual research cards.
Correction policy: When a factual issue or source discrepancy is identified, we will review the evidence and correct the research page. Evidence should be cited with its source title, observation date and link to the original material.