SEOAIO.aiSearch Engine Optimization + AI Optimization

Fact Sheet: tell us what is true about you

Record your ground-truth facts so we can measure when an AI engine states something wrong about your business.

What it is

Your fact sheet is a short list of things that are true about your business, each paired with the wrong answers we should watch for. It is the reference Answer Accuracy measures against, which is what turns “the AI got us wrong” from an impression into a measurement.

When to use it

Set it up once, early. Add a fact any time you notice an AI assistant stating something incorrect about you — the wrong founding year, the wrong category, a service you do not offer, a location you left years ago.

Step by step

  1. 1Open Fact Sheet from the sidebar, under Action Queue.
  2. 2Name the fact in your own words, for example “founding year”.
  3. 3Enter the correct value, plus any other acceptable phrasings (comma separated) so a right answer worded differently is not counted as a contradiction.
  4. 4Enter the wrong values to watch for, comma separated. This field is required.
  5. 5Add the fact. It applies from your next measured run onward — we never backfill it onto earlier runs.
  6. 6Retire a fact when it stops being true. Retiring stops future evaluation and keeps the record.

What you will see

Your active facts, each showing the correct value, its accepted phrasings, and the wrong values being watched. Once a measured run has sampled the engines, the Answer Accuracy side of the two-sided AI panel on Analyze stops reading “not yet measured” and starts reporting contradictions, each with the engine’s own quote as evidence.

Measured, never manufactured

Wrong values are required, and that is deliberate. Without at least one there is nothing measurable to detect, and the only alternative would be asking a language model whether another language model sounded wrong — a model’s opinion, not a measurement. Retiring a fact never deletes it, because past accuracy runs were evaluated against it and deleting would silently change what an old measurement meant.

Next

Answer accuracy: when AI gets you wrong

The two-sided view of AI visibility: presence gained on one side, incorrect-mention risk on the other.