A buyer asks ChatGPT "Best [category] vendors" — and your brand isn't in the answer. Or worse, it's named once, in a footnote, after three competitors you've outsold for a decade.
There are typically four reasons. Most brands assume the wrong one.
Reason 1 — The authority stack is too thin
Often the most common cause. The brand exists. The brand has customers. The brand has revenue. But the citation graph the model retrieves from doesn't have enough authoritative sources naming the brand for the category prompt.
Press releases on the brand's own site count less than the brand typically assumes. Pay-to-play industry rankings count less than the brand typically assumes. Tier-1 earned media — Reuters, Bloomberg, The Wall Street Journal, The New York Times, Forbes, TechCrunch, Wired — tends to count the most. So does Wikipedia. So does original research the engines can cite.
If the brand has a thin tier-1 footprint, the model often has little to retrieve.
Reason 2 — The brand is cited, but for the wrong prompt
The brand shows up when the prompt names the company directly — but disappears when the prompt names the category. That gap often means the model knows the brand exists but doesn't associate it with the category at the level required to surface it.
This is a prompt-to-entity association problem. The fix is typically repetition: get the brand named alongside the category in trusted sources, at volume.
Reason 3 — A negative or stale narrative is dominating
The brand is in the citation graph — but the model's compressed summary is unfavorable. Layoffs from two years ago. A regulatory issue from 2022. A founder controversy that resolved but never got the resolution covered.
The model is doing its job. The narrative input is stale.
Reason 4 — The brand is new to the category
Sometimes the diagnosis is simple. The brand is two years old. The category prompt favors incumbents the model has seen for a decade. Time and tier-1 coverage tend to be the primary fixes.
How to diagnose which one applies
Run the brand through a structured audit across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews. Test the brand name directly. Test the category prompt. Test five competitor comparison prompts. The patterns usually surface fast. [Read: How to Audit What Every Major AI Engine Says About Your Brand]
The audit tells you which of the four reasons is yours. The reason tells you the repair plan.
What doesn't work
Buying ads on the engines — the retrieval layer generally doesn't pull from ad inventory. Filing a complaint — there is no "claim my listing" for AI answers. Waiting for the next training cycle — most engines now update retrievals on shorter cycles.
What tends to work
Tier-1 earned media on the category prompt. Wikipedia. Structured owned authority. Original research the model can cite. The discipline that produces those is Generative Engine Optimization — GEO — and it's now its own function. [Read: AI Communications · Signals That Move AI Reputation]
See also: Signals That Move AI Reputation · AI Reputation Glossary
No communications firm can guarantee specific outputs inside third-party AI systems. The discipline is shaping the inputs the engines retrieve from — not directing the engines themselves.
