What is GEO / AEO (generative engine optimization)?

How it differs from SEO

SEO competes for a position in a list. Ten results appear, and being fourth still earns traffic.

An assistant does not produce a list. It produces an answer that names two or three options, and everything not named is absent rather than lower down. There is no page two to be on. That single structural difference drives most of what makes GEO a distinct discipline: the payoff curve is far steeper, and the mechanism that decides inclusion is not a ranking function you can reverse-engineer from position data.

The second difference is what the system is optimising for. A search engine ranks documents. A model synthesises an answer from what it has internalised in training and what it retrieves at query time — see trust graph. Being retrievable and being believed are different problems.

What actually moves it

• Third-party corroboration. What independent sources say about you carries far more weight than what you say about yourself. Review sites, industry publications, comparison articles and community discussion are disproportionately influential.
• Consistency across sources. A model reconciling contradictory claims about your pricing or positioning will hedge or omit you. Inconsistency reads as unreliability.
• Machine-readable specificity. Concrete, checkable facts — what it does, who it is for, what it costs, what it does not do — outperform positioning language that a model cannot verify or repeat.
• Structured data and clean retrieval. If a crawler cannot parse your claims, they will not be retrieved at query time.
• Being the answer to a question, not a page about a topic. Content organised around the questions buyers actually ask is what gets lifted.

What does not work

Two categories of effort are being sold and are mostly wasted.

Keyword density and volume plays. Publishing more pages about a term does not make a model recommend you. The stub-page glossary, the thin comparison page and the AI-generated content farm all fail here for the same reason they now fail in search: nothing about them constitutes evidence.

Anything promising guaranteed placement in organic answers. Inclusion in an answer is either earned or paid and disclosed. A vendor offering guaranteed unpaid placement is either describing paid placement or describing nothing.

How to measure it

Presence is measured by sampling, not by a rank tracker. Fix a prompt set that reflects real buyer language, run it repeatedly across the assistants that matter, and record whether you appear, in what position and in what framing. That is AI share of voice, and framing matters as much as presence — being named as the cheap option in a premium comparison is a mention that costs you.

GEO and paid answer insertion move the same business outcome and are best run as one programme: paid presence works immediately and stops when spend stops, organic authority compounds and cannot be bought directly.