← how ARIA works
Stage 03 · representation

Whether AI describes you accurately

Citation asks whether engines talk about you. Representation asks what they say — is it true, and how does it frame you? A confidently wrong answer about your price, product, or compliance loses deals, so I check the answers engines give against your own pages.

How I check

It rides the same probe as Citation — no extra schedule, no setup, nothing for you to fill in. For every answer that names your business, I compare its claims against your crawled pages — your own site is the source of truth — and read the overall framing.

Quote or stay silent

I only flag a claim as wrong when I can quote the exact sentence on your site it contradicts. If your pages don't address a claim, I say nothing — I never guess that an engine is wrong about you. That keeps a false “the AI got you wrong” flag close to impossible.

Accuracy & framing

Two readings each pass, both from answers that mention you:

Contradictions
A claim in the answer that conflicts with a verbatim quote from your site — shown as “AI says X, your page says Y.” These are stale or wrong beliefs to correct.
Sentiment
How the answer frames you — positive, neutral, or negative — averaged across the answers that name you.

Claims your site simply doesn't address are left alone, not surfaced as questions — that keeps the card to things you can act on with confidence.

What to do with it

Your page is the authority, so the fix is always to make the truth easier for engines to learn: correct or strengthen the page, add structured data, and earn citations so the model re-learns it. Sentiment is a monitor — a trend to watch, not a number to chase.

No score, on purpose

Unlike Readiness and Citation, Representation has no 0–100 and never feeds your presence score. What an engine believes about you isn't something you move directly or on demand, so I show it as a monitor and a fix list, not a grade. Honest beats precise.