Generative Engine Optimization, or GEO, gets described either as "SEO is dead" or as "nothing new, just do good SEO". Both are wrong. A lot carries over, but the goal, the unit of success and the way you measure it all change.
The goal: from a click to a mention
Classic SEO optimizes for a position in a list of links, and success is a click. In an AI answer there is often no list to rank in. There is a paragraph that names two or three options, sometimes with sources, sometimes without. Success is being one of the names, described correctly, ideally with a link.
That is a different shape of outcome. You can be cited without getting a click, and you can lose a sale without ever seeing the impression.
The unit: from a keyword to a question
People type short keywords into a search box. They type full, specific questions into an assistant: "what's a good CRM for a five person agency that already uses Google Workspace?". Small changes in phrasing can change which brands come back.
So instead of tracking a few hundred keywords, you track a set of realistic questions, grouped by intent. The question set becomes your main asset.
What carries over from SEO
- Crawlability. If a crawler cannot fetch your page, no engine can quote it. Server-rendered content, clean HTML and sensible robots rules still matter.- Clear structure. Headings, short paragraphs and lists help retrieval systems extract the right passage.- Authority and mentions. Being referenced by trusted, relevant sources still helps, perhaps more than ever, because models echo what many sources agree on.- Answering the question. Pages that genuinely explain something keep winning.## What changes ### Passages matter more than pages Retrieval-based engines often pull a passage, not a whole page. A page that states its answer in the first two sentences of a section is easier to quote than one that builds up to it over six paragraphs. ### Entity clarity matters a lot Models need to understand what your brand is and what it does. If your site, your social profiles and the directories that list you describe you in different ways, the model's picture of you gets blurry. Repeating a consistent one-line description of your brand helps. ### Off-site content carries more weight Forums, community threads, comparison articles, review platforms and videos regularly appear among the sources of AI answers. Your own website is only one of the places that shape how you are described. ### Each engine behaves differently Some engines cite sources inline, some rarely do, some lean on fresh web results, some lean on what the model already learned. A strategy tuned for one assistant does not automatically transfer to another, which is why measuring across several engines matters. ## Measuring it Rankings do not translate to this world. The useful metrics look more like this:
- Mention rate: on how many of your target questions are you named?- Share of voice: how often are you named compared with competitors?- Accuracy: is what the engine says about you true?- Cited sources: which pages are behind the answers, yours or someone else's?Tracking those by hand works for a first audit. Over time you want them on a schedule, across engines, with history, which is what tools like CiteMe do. ## The short version SEO taught us to be findable. GEO asks us to be quotable. Keep the technical basics, write pages that answer one question clearly, be talked about where models read, and measure mentions instead of positions.
Top comments (0)