Why Your Competitor Appears in ChatGPT and You Do Not

You ask ChatGPT the question your buyers ask and a competitor comes back named. You do not. The instinct is to assume they out-spent you or out-ranked you. In our measurements neither is usually the cause. There are four reasons, they are distinguishable in about 20 minutes, and they take four different fixes with very different price tags.

The short version

We ran 8 Dubai buyer questions through Google AI Mode on three dates and separated the causes. Cause one: the engine cannot resolve who you are, so you are not a candidate. Cause two: it knows you but has no reason to prefer you. Cause three: your pages are not retrievable at the passage level.

Cause four is the most common in our data: the answer is assembled from surfaces you do not own. On our 62-query panel (16-18 September 2026) the four platforms hold 18% of citations, more than any single website, and 27 of the 56 answers that cited anything cited one of them. Our own domain holds 1.7%. Only the third cause is fixed by writing more pages on your own site.

The four causes, and how to tell them apart

Cause 1: the engine cannot resolve you

Ask the engine directly who you are. If it confuses you with a similarly named company, invents a description, or returns nothing, you have an Entity problem and nothing downstream will work until it is fixed. We hit this ourselves: the acronym AEO also denotes a customs status and a US clothing retailer, which polluted our own query research until we filtered it.

Cause 2: it knows you but has no reason to prefer you

The engine describes you correctly and still names someone else. That is a Voice and distinctiveness problem. Our July panel caught the shape of it: Digital Nexa was named in 37.5% of answers with a 0% citation share, meaning the engine preferred a brand whose pages it had not even used that day.

Cause 3: your pages are not retrievable

You are described correctly, you are distinct, and your pages still never appear in the source list. That is passage-level structure. The Princeton GEO research, Aggarwal et al., measured passage changes of this kind moving visibility by up to 40% in controlled tests.

Cause 4: the answer is not sourced from anyone's website

Read the source list under the answer. If it is Reddit threads, YouTube videos and third-party lists, your competitor did not beat you on-domain. They are present where the engine is looking, and you are not.

The diagnostic

What you observe	Cause	The fix	Typical cost shape
Engine gets your identity wrong	Entity unresolved	Entity clarity, consistent descriptions, structured data	Low cost, fast
Described correctly, rival named	No reason to prefer you	Positioning, proof, a claim only you can make	Slow, highest value
Never in the source list	Passage structure	Answer blocks, quantified claims, inline citations	Low cost, fast
Sources are platforms, not sites	Off-domain answer	Placement and participation on those surfaces	Ongoing, rarely staffed

Run the diagnostic yourself

1. Ask the engine who you are. Resolution problem or not, you learn it in one query.
2. Ask your 8 buyer questions and record whether you are named, cited, or absent for each.
3. Read the source list under every answer, and classify each source as a competitor site or a platform.
4. Count the split. If most sources are platforms, stop planning on-domain work for those questions.
5. Repeat in 60 days, because a single reading in a volatile system tells you nothing about direction.

What our own diagnostic returned

One number is worth holding on to before you blame a rival: no single domain in that panel held more than 6% of the citations, and the largest non-platform source held 2%. The competitor you are watching did not take the answer. The answer was assembled from a crowd you are not part of.

We run this on ourselves

We run this on ourselves and publish the result. Across those 8 Dubai questions, Ivanooo scored 0% named and 0% cited on 21 July, 7 September and 12 September. Across the wider 62-query panel our Share of Recommendation reads 0.0, with 59 of 62 questions returning us absent. Our diagnosis is cause four with a side of cause three: the answers we want are heavily platform-sourced, and our own pages were not built at passage level until recently. As Firoz Azees puts it: "a measured zero is a work order; a flattering estimate is a story you tell yourself."

Questions buyers ask

Could my competitor simply be paying for it? Not for the organic answer. Generated answers are assembled from retrieved sources, and there is no slot to buy in the mechanism we measured.

They have more backlinks. Is that the reason? It can help them appear in the underlying pool. It does not explain being named while uncited, which our panel recorded repeatedly, and it is not a fix you can act on this quarter.

How long until I appear after fixing the cause? Longer than any vendor will promise. We have held the method on ourselves for 52 days across three readings without winning these questions, which is the honest benchmark.

Does the engine punish me for something? Nothing in our data suggests punishment. Absence is a selection outcome, not a penalty, which is why it responds to evidence rather than appeals.

If the answer is platform-sourced, should I abandon the question? No. You change the instrument. Being present on the cited surface is the work, and the limits of on-domain AEO explain why nothing else reaches it.

Can I fix more than one cause at a time? Yes, and the cheap two should go first. Entity clarity and passage structure are fast and low cost. Distinctiveness and placement are slow and expensive.

How do I know the fix worked? Only a re-probe tells you. The full method is in how to verify any AI-visibility claim, and it applies to your own work as much as to a vendor's.

At Ivanooo, Firoz Azees runs Distinctiveness Engineering for the AI-answer era: measuring who the engines name, cite and recommend, then engineering the gap between listed and chosen. Start with a free AI visibility check and find out which of the four causes is yours.