AI Visibility, GEO, AEO, LLM SEO: an iGaming Glossary That Actually Helps

Every few months the industry invents a new acronym for the same underlying shift: players ask AI assistants instead of search engines, and brands win or lose inside answers they never see. The vocabulary around this shift is genuinely confusing, partly because agencies coin terms faster than anyone can define them. This page is our working glossary. We use these definitions internally at FoeGlass, and we will keep the page updated as the language settles.

AI visibility

How often, how prominently, and in what tone AI models mention your brand when a relevant question is asked. It is the umbrella metric. A brand can rank first on Google and still be invisible in ChatGPT answers, because models draw on different sources, weigh them differently, and add their own safety layer on top. AI visibility is measurable: you ask models real player questions, many times, across markets, and you record what comes back. That measurement is literally what the FoeGlass index is.

GEO, generative engine optimization

The practice of improving how generative engines describe and recommend you. The name mirrors SEO on purpose. The difference is the unit of competition. In SEO you compete for a ranked list of links. In GEO you compete for a sentence inside a synthesized answer, and there are usually three to seven brand slots in that sentence. GEO work includes getting cited by the sources models trust, keeping licensing facts unambiguous, and fixing the specific phrases where models describe you incorrectly.

AEO, answer engine optimization

A close sibling of GEO, focused on answer surfaces: Google AI Overviews, Bing Copilot, Perplexity's cited answers. AEO leans harder on structured data, crawlability and question-shaped content, because these surfaces quote and link their sources. In practice most teams treat AEO and GEO as one discipline, and honestly, so do we.

LLM SEO

An informal umbrella people type into search when they first meet the topic. It usually means GEO plus AEO plus some prompt-testing. If someone sells you "LLM SEO", ask what they measure. If the answer does not include repeated sampling across geos and models, they are guessing.

Deposit leak

Our term for a specific, costly failure: an AI answer in a regulated market points the player to a brand that is not licensed there. The player deposits somewhere the local regulator does not protect, and a licensed operator quietly loses a customer it paid taxes to be allowed to serve. We wrote a full breakdown of the mechanics in a separate post.

Game-Bridge

The route from a game question to a casino recommendation. Players often ask about a slot or a crash game, not a casino. The model answers the game question and then names venues where you can play it. That second half of the answer moves real deposits, and casinos rarely monitor it. Our Game-Bridge board tracks which games route to which casinos.

Measurement cycle

One complete pass of the index: every tracked question, in every market and language, on every engine, with repeats. We publish results per cycle rather than continuously, because models change in steps and a dated snapshot is easier to reason about than a rolling average. Scores compare across cycles, which is how movement becomes visible.

Refusal

The engine declines to answer a gambling question. Refusal policies differ by engine far more than most people expect, and the spread is wide enough to change how you read any per-engine result. A refused answer contains no brands at all, which has strategic consequences we cover in a dedicated post.

Verbatim

The exact answer text as captured, stored with its date, engine, market and language. Screenshots age badly and are easy to doubt. A verbatim with metadata is evidence. Every score in the FoeGlass index drills down to the verbatims behind it, because a number you cannot verify is just an opinion with decimals.

If a term you keep meeting is missing here, write to us and we will add it. The glossary works best when it answers real confusion, not imagined confusion.