SolCrys Logo

05 Diagnose · 7 min

When presence is not the problem

Two failure states that a presence metric will never show you: the engine describes you wrongly, or describes you correctly and still picks someone else.

Being present and wrong

An answer that mentions you and misdescribes you is worse than one that omits you, because it actively teaches a buyer something false and it does so with the engine's authority behind it. Presence metrics score this as a win.

Grading requires the reference document from Module 2, which is why context came first. The frequent finding is not dramatic hallucination but quiet staleness — a retired price, a feature you sunset, a positioning you moved away from two years ago. The web still says it, so the engine still says it.

Note the coverage limit: anything your context is silent about cannot be graded. If accuracy grading comes back clean, check whether that means you are accurate or whether it means your context does not cover the topics where you are wrong.

Being present, correct, and still not chosen

The hardest state. The engine has your facts right, cites your page, and recommends a competitor anyway. There is nothing to correct.

This is almost always corroboration. The competitor's claim is echoed across independent sources; yours exists on your own site and nowhere else. The engine is doing what Module 1 described — favouring what several sources agree on — and the fix is not a better page. It is a third-party source that says the same thing.

That is slower and less controllable than editing your own content, which is why teams avoid diagnosing it correctly and ship another landing page instead. If you are in this state, Module 5 has to be about the source layer, not about your own site.

Where this lives in SolCrys

  • Answer Accuracy grades responses against your Corporate Context and flags divergences.
  • The grounding view shows what the grade was based on, which is how you tell a real pass from a coverage gap.
  • Recommendation is scored per response, so you can find responses that are accurate and still unfavourable.

Check yourself

  • Is there anything an engine says about you that is stale rather than false?
  • If accuracy grading looks clean, how would you tell whether that is real or a coverage gap?
  • If you are cited, accurate, and still not recommended, what does that tell you the fix has to be?

Go deeper

This lesson is the spine. These guides are the depth.

Risk Monitoring

When AI Describes Your Brand, Is It Telling the Truth?

AI answers drop claims you earned, quote prices you retired, and assert things you never said, and that text can be steered on purpose. How to grade every AI answer against your own grounding truth, with receipts.

Risk Monitoring

How to Fix a Wrong Fact in an AI Answer About Your Brand

An AI engine is stating a wrong price, a dead feature, or a bad comparison about you, and there's no one to email. The wrong fact is a relayed source. Here's how to find it, fix it or outweigh it, and re-test until the answer flips.

Measurement

AI Recommendation Score

AI can name your brand and recommend a rival in the next sentence. The Recommendation Score grades every AI answer 0-100 on how favorably it positions you across ChatGPT, Gemini, Perplexity, Google AI Overviews and Claude, plots you against every competitor, and shows the verbatim line behind every point.

Continue

Free · No credit card

Turn AI answer gaps into governed marketing execution.

Start free with a ChatGPT visibility read, then add multi-engine tracking, Corporate Context governance, and the action-to-result loop when you are ready.

Start Free