A Wikidata entry and 6 weeks later: brand started appearing in AI answers

A Wikidata entry and 6 weeks later: brand started appearing in AI answers

Question: do structured entity records (Wikidata) influence whether large language models name your brand? One before/after probe.

What was done: a niche tool with no Wikidata presence got a properly-sourced entry (with statements linking it to its category and founders) plus matching sameAs schema. I then re-ran a fixed set of 25 category prompts across three AI assistants weekly.

Outcome: pre-intervention, the brand was named in 1 of 25 prompts on average across assistants. Six weeks after the Wikidata entry, that rose to 6 of 25. Competitors with longstanding Wikidata entries averaged 9 of 25 over the same period.

Caveat: AI assistants update via opaque, irregular retraining and retrieval, so timing-based attribution is weak; the brand also earned a few press mentions in the window. n=1 brand, 3 assistants.

Method note: fixed 25-prompt battery, weekly runs across three assistants, single brand pre/post.

Confidence: low — suggestive and theory-consistent, but uncontrolled and timing-confounded.