What should make an AI visibility score unpublishable?

by

Most analytics products are designed to always return a number. For AI-search visibility, I think a trustworthy product needs the opposite safeguard: the ability to withhold a score when the underlying evidence is not decision-grade.

At Corank, I am testing five publication gates:

1. Traceability — every chart point must lead back to the exact prompt, raw model answer, citations, timestamp, collection status, and methodology version.

2. Denominator integrity — failed and excluded runs must remain visible. A timeout cannot silently become “brand absent,” and an ineligible observation cannot quietly disappear.

3. Replayability — the stored answer should reproduce the original classification. A fresh model response is useful for comparison, but it cannot replace historical evidence.

4. URL lineage — raw, redirected, resolved, and canonical URLs should be connected with an explicit reason. Cleaning data must not erase the citation that was actually observed.

5. Counterfactual behavior — removing a brand mention, changing completion state, or expiring stale evidence should change the relevant metric in a predictable way.

The product decision is not just pass or fail. It can be: publish normally, publish with a reliability warning, show components only, hold the last valid score, or suppress the score until evidence is repaired.

I wrote the full falsification protocol here:

Corank applies this evidence-first approach to AI-search visibility:

For other founders building analytics products: what evidence failure would make you suppress a score instead of merely adding a caveat?

2 views

Add a comment

Replies

Be the first to comment