Quality and evidence · Branch Q.C, Originality and defensible value
Q-10 Research and benchmark methodology must be understandable
Test conditions, sample selection, version, number of repetitions, measurement method, conflict of interest and limitations must be explained well enough for the result to be evaluated.
- Scope
- All public content
- Priority
- High priority.
- Quality bucket
- Original, non-commodity value
- Legacy coverage
- GEO-22
Why it matters
A benchmark result with no method behind it is just a claim, and readers cannot tell whether it is fair or repeatable. Search engines and AI answer engines have no basis for trusting a result they cannot assess. Hidden sponsorship or missing limitations make that worse.
How to implement
- Add the test date and scope to the page.
- List the hardware, software and model versions used.
- Describe the input dataset or task set, the number of repetitions and the statistic used.
- Disclose any sponsorship or conflict of interest.
- Add a limitations section, then ask someone independent to read the method and confirm they can follow how the result was produced.
Checklist
- State the test date and scope.
- State the hardware, software and model versions.
- Explain the key characteristics of the input dataset or task set.
- State how many repetitions were run and which statistic was used.
- Disclose any sponsorship or conflict of interest.
- Add a limitations section.
How to audit
- Check whether an independent reader could understand from the method how the result was produced.
How UltimateSEO checks this automatically
This control needs human evidence: the published methodology for any research, benchmark or test and whether it can be followed by a reader.
States used: PASS, FAIL, UNKNOWN, NOT_APPLICABLE or BLOCKED. UNKNOWN is never shown as a pass. How states work.
Sources
Part of The UltimateSEO Framework v4.1, revised 2026-10-07. Evidence states, control states and the publish gate are explained on the hub.