Generative engine optimization: a plain guide

What generative engine optimization (GEO) means, how it relates to AEO and SEO, and the five steps that get a B2B brand named in AI answers, in plain words.

On this page
  1. What is generative engine optimization?
  2. How is GEO different from SEO?
  3. Step 1: Choose the questions your buyers ask
  4. Step 2: Measure where you stand
  5. Step 3: Find the questions you lose
  6. Step 4: Publish pages that answer them
  7. Step 5: Get onto the pages the engines read
  8. Then repeat, every week
  9. What GEO is not
  10. Common questions

Generative engine optimization, or GEO, is the work of getting your brand named in the answers AI engines write. When a buyer asks ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews or Google AI Mode a question, GEO is what decides whether your name is in the reply.

The term is new, and so is a lot of the advice around it. This guide keeps to the plain version: what GEO is, how it fits with what you already do, and the steps that work.

What is generative engine optimization?

A generative engine is an AI system that writes an answer rather than returning a list of links. Optimizing for one means shaping what it says about your category so that it names you, names you early and describes you well.

You will see a few names for the same work:

They all point at one goal. This guide uses GEO for the work and AI visibility for the result.

How is GEO different from SEO?

SEO aims to rank a page. GEO aims to get a brand named in an answer. Much of the groundwork is shared: clear pages, a site engines can read, other sites that mention you. What changes is the target and the way you measure it. Our post AI visibility vs SEO goes through what carries over and what is new.

Step 1: Choose the questions your buyers ask

GEO starts with questions, not keywords. Write down the questions a buyer asks on the way to choosing a product like yours:

Aim for a set wide enough to cover your main use cases and narrow enough to stay about buying. Our post Questions to track in AI answers shows how to build and group the set.

Draft questions, for you to approve
Category

What is the best payroll software for a 20-person company?

Category

Which payroll software handles tips for restaurant staff?

Comparison

Is Tessera or Novaly better for a restaurant group?

Comparison

What are the best alternatives to Novaly?

Problem

How do I pay staff who work in more than one state?

Example, using an invented brand

Step 2: Measure where you stand

Ask every question on every engine you care about, and record which brands each answer names, in what order, how it describes them and which pages it cites. Then count.

Five measures cover it, read in this order:

  1. Mention rate: of the answers read, the share that named you.
  2. Share of voice: your share of all the brand mentions in the answers.
  3. Average rank: your place among the brands named, when you are named.
  4. Your pages cited: the share of cited links that point to your own pages.
  5. Sentiment: how warmly the answers describe you.

Answers vary from one asking to the next, so count across many answers and repeat the same set every week. That turns a pile of replies into figures you can compare.

Step 3: Find the questions you lose

A question you lose is one where a rival is named ahead of you, or named while you are not. Group them by topic. The topics where you lose the most questions are where your first pages should go.

Read what the answers say about the rival on those questions. The reasons an engine gives for naming them are the points your own pages need to answer.

Step 4: Publish pages that answer them

For each topic you lose, write one clear page that answers its questions.

Step 5: Get onto the pages the engines read

Engines that search the web lean on review sites, comparison articles, community threads and partner pages. Find the pages they cite for the questions you lose, then do what each one calls for:

This is often the fastest step. A page that already ranks and already gets cited can bring your name into answers soon after you are on it.

Then repeat, every week

GEO is a loop. Ask the same questions each week, see what moved, and pick the next page to write or place to get onto. When you publish a page, note the date, so you can see whether its questions start naming you in the weeks after.

What GEO is not

A few things often sold as GEO do not hold up:

Common questions

Is GEO only for big brands?

No. Specific questions favor brands with a clear fit. A smaller brand that answers a narrow question well can be named on it before a larger brand that never addresses it.

How long does GEO take to work?

Getting onto pages the engines already cite can show up within weeks on the engines that search. New pages of your own take longer to be found. What the underlying models know changes slowest, when they are retrained.

Which engines should I track?

The ones your buyers use. In B2B that usually means all six major ones, because each answers differently and buyers move between them.

What you gain with BotSay

  • Every step of the loop in one place: your questions asked on six engines, every week.
  • The five measures, counted from the same answers, so you see where you stand and what moved.
  • The questions you lose, grouped by topic, with a brief for the page to write.
  • The pages the engines cite, with what to do about each one.
  • One fixed monthly price with everything included and no add-ons. Plans differ only in the number of questions, from $59 a month for 60 questions.

Keep reading

Your first results arrive right after setup

Find the page to write next.

Monthly plans from $59. Cancel at any time.

Get started