TL;DR
SEO ranks pages in a list. AEO (Answer Engine Optimisation) gets your content surfaced as the answer in features like Google's AI Overview, Bing Copilot results, and voice assistants. GEO (Generative Engine Optimisation) ensures your brand is cited inside generative-AI answers — ChatGPT, Gemini, Perplexity, Claude.
They're not replacements for SEO. They're additional surfaces that share most of the same fundamentals (authority, structure, citations, clarity) and diverge meaningfully on measurement and tactics.
"If a brand isn't earning citations inside the AI answer, it isn't in the consideration set anymore. The list ranking still matters — it's just no longer the only thing that does."
Why this matters in 2026
Three numbers we've been tracking across our retainer book over the last twelve months:
- Average zero-click rate on informational queries: up 28% YoY in our portfolio.
- Average traffic per ranking #1 result: down 14% YoY for non-branded queries.
- Brand mentions inside ChatGPT/Gemini answers (where we measure): up 4.6× for clients running structured GEO programs vs control accounts.
Translation: the SERP is getting more crowded with summarised answers, and the AI assistants are getting more confident at giving recommendations. The brands that show up in both places win the customer mindshare. The brands that only show up in classical search are slowly getting outflanked.
The stack, defined properly
SEO — classical search visibility
Still the foundation. Ranking pages in the blue-link list, the local pack, the image carousel, the video carousel, the featured snippet. Driven by relevance, authority, technical health, and user-experience signals. It's not going anywhere — it's just no longer the whole picture.
AEO — Answer Engine Optimisation
Optimising content to be surfaced as the answer in features that synthesise an answer rather than show a list. Examples include:
- Google's AI Overview (formerly SGE)
- Bing Copilot in-SERP answers
- Apple Spotlight intelligence
- Voice-assistant responses (Alexa, Siri, Google Assistant)
AEO is mostly an evolution of structured-data and featured-snippet practice — but with much higher precision required around the structure, the claim, and the citation.
GEO — Generative Engine Optimisation
Optimising for the conversational AI assistants that don't have an open-web-style SERP at all: ChatGPT, Gemini, Perplexity, Claude, Meta AI, Copilot Chat. Here the unit isn't a ranking — it's a citation or a recommendation inside the model's answer.
GEO requires three things classical SEO doesn't fully require:
- Authoritative entity presence — Wikipedia, Wikidata, industry citations.
- Citable, quotable content — explicit statements of fact and opinion the model can quote with confidence.
- Wide reference corpus — your brand mentioned in third-party content the models trained on or retrieve from.
How to measure (without vibes)
The biggest difference between agencies who can do this work and agencies who claim they can is whether they have a reproducible measurement protocol. Here's the one we use:
1. Prompt suite
For each client, we develop a curated suite of 50–200 prompts covering: branded recall, comparative ("best X for Y"), problem-discovery ("how do I…"), and recommendation prompts. Suite is locked at quarter-start and run on a fixed cadence.
2. Sampling protocol
Each prompt is run with controlled variations (cold session, account-logged-in session, geographic variant) across all major models, three times per cycle, with results captured to a structured database. Yes, it's tedious. Yes, it's the difference between data and vibes.
3. Reporting
We report on three numbers monthly:
- Citation share: % of prompts in which your brand is mentioned at all.
- Recommendation share: % of "best X for Y" prompts in which your brand is the (or a) recommendation.
- Sentiment & accuracy: how the brand is being represented and whether the model is hallucinating about you.
What we'd do on Monday
If you've read this far, you probably want to know what to actually do. Here's a 30-day starter that we run with every new GEO retainer:
- Week 1: Establish your prompt suite and run baseline. You can't manage what you can't measure.
- Week 2: Audit your top-50 pages for citability — single-claim sentences, original data, quotable statements, clear authorship.
- Week 3: Audit your entity presence — Wikipedia/Wikidata, industry directories, schema.org coverage, knowledge graph hygiene.
- Week 4: Build the 90-day plan. Three workstreams: content rewrites for citability, off-site authority, structured data depth.
Boring? Maybe. Effective? Across the eleven retainer clients we've run this exact playbook with, average citation share improved from 6% → 38% in 90 days. The fundamentals work. They always have.
FAQ
Is GEO really a separate discipline, or just SEO with new tactics?
Both. The fundamentals are the same — authority, clarity, structure, citations. The tactics, measurement and surfaces are different enough that pretending it's "just SEO" is going to leave value on the table.
Should I be worried about being cited but not clicked?
Yes and no. Cited-but-not-clicked is real and growing. But brand mention inside an AI answer is a powerful credibility signal — it directly affects consideration and indirectly affects branded search. We measure both effects.
What about llms.txt and bot blocking?
Big topic, separate post coming. The short version: most B2B brands should be visible to most major AI crawlers. The economics for blocking are weaker than the economics for being cited. There are exceptions.
Have a brief that touches any of this? Send it through. We'll reply within three working days with a quick read on what we'd do — for free, no discovery-call gymnastics.