How we rate the receipts

Most published “research wins” are case studies written to make someone look good. That doesn’t make them false. It does mean you should know what you’re looking at before you cite one in a budget meeting.

So every entry gets two labels: how strong the evidence is, and who’s telling the story.

Evidence strength

TestedControlled experiment or A/B test tied to the research-driven change.
MeasuredBefore/after metrics, but no control group.
ReportedThe company states an outcome but doesn't share how it was measured.
DecisionResearch changed a decision (killed, pivoted, re-prioritized). No metric.

When a case sits between two levels, we pick the lower one. “Decision” isn’t a weaker category. It’s a different kind of win: the expensive thing that didn’t get built. Those rarely come with a number, and they’re often research’s biggest payoff.

Source type

Company's own accountThe company's own blog, talk or report.
Agency / vendorA consultancy or tool vendor writing about its own client work. Has an incentive.
IndependentJournalism, academic work, audits, books.
Community submissionSubmitted to Receipts by a practitioner.

Attribution

A redesign that improved conversion isn’t automatically a research win. Each entry spells out what the research found, what changed because of it, and how directly the source ties the outcome to the research. If the link is loose, the Attribution and Caveats sections say so.

Ground rules

  • Every seed entry links to at least one public source we opened and checked.
  • Entries are summaries in our own words. Go to the source for the full story.
  • Numbers are quoted exactly as the source reports them. We don’t extrapolate.
  • Community submissions are reviewed before they appear. Anonymized entries are allowed.

Found a mistake?

Open an issue with the correction and a source. Or submit a win of your own.