Preparing the sample workspace…
Evidence method
Ekvidu treats every result as a labelled observation with provenance, coverage and limits, not as a universal score for how every AI product sees a brand.
02 / Deliverables
Frozen benchmark cells and explicit denominators
Raw evidence lineage, timestamps and collection-status reasons
Separate mention, citation, source-support and traffic measures
Compatibility checks for baseline and follow-up comparisons
03 / How it works
Every collection route must be permitted for the intended use and labelled by provider, model or declared benchmark, access method, search or grounding state, locale and time. API output keeps the label API benchmark until matched evidence supports a narrower consumer-surface claim.
The denominator includes the planned panel. Successful, invalid, retried, cancelled and not-run cells remain distinct. A result is interpreted only with its coverage, so an outage or refusal cannot masquerade as a brand absence.
A strict follow-up uses fresh evidence under compatible prompt, benchmark, market and attempt conditions. Mismatched explorations may be useful for discovery, but they do not support a causal claim or a guaranteed outcome.
04 / Limitations
05 / Next step