GEO explained: how to get mentioned in AI answers

Three company names in a four sentence answer

Imagine a machine builder looking for a supplier of custom gearboxes. Around the table sits the DMU, the decision-making unit: a buyer, an engineer, someone from quality and a director who guards the budget. At some point that week, each of them asks ChatGPT something. The buyer wants to know which suppliers are worth a request for quotation, the engineer which parties can meet those tolerances.

Every time, within a few seconds, there is an answer with three company names. No list of twenty links, no second page and no cup of coffee halfway through scrolling. Three names.

If your company shows up for several people in the DMU, your name is already on the table before anyone has called you. If it does not, you will never hear about it, because nobody calls to say they could not find you.

That is what GEO is about, along with AEO, the term that keeps appearing next to it. In this blog you will read what they are, where we stand today and what you can do about it this week.

GEO: writing for a reader who has read everything

GEO, Generative Engine Optimization, is the work that makes AI answer engines use your content and mention your company. Think of ChatGPT, Perplexity, Gemini and the AI summaries at the top of Google.

The term comes from a study by Pranjal Aggarwal and colleagues, presented at the KDD conference in 2024 (GEO: Generative Engine Optimization). The researchers tested which changes to a text increase the chance that an AI uses that text in its answer. Three things worked best: citing sources, adding quotations and adding statistics. These gave a relative improvement of 30 to 40 percent on their visibility measure. Fluent, clear writing scored 15 to 30 percent better. And repeating keywords, the old SEO reflex? Little to nothing. A text that says 'gearbox supplier' ten times does not impress an AI. It remains an academic experiment, so read the percentages as a direction.

Classic SEO is about a spot in a list of links. With GEO that list is gone, because the user gets one composed answer with a few sources at most. Your content needs to be a good building block for that answer, and your brand needs enough authority to be mentioned. You are really writing for a reader who has read everything, never gets tired and sums up the answer in four sentences.

AEO: what a good answer looks like

AEO, Answer Engine Optimization, is older than the AI wave. It emerged with featured snippets, the 'people also ask' boxes and voice assistants: the small box at the top of Google where your question was already answered. Anyone who wanted to appear there had to answer a concrete question in two or three sentences, directly below a heading that asks that question.

GEO and AEO largely overlap. A useful division: AEO is about the form of your content. Are there clear questions and short answers on the page, is the structure clear and does a machine understand what is there thanks to schema markup? GEO is about trust in the source. Is your company mentioned elsewhere too, is your information consistent everywhere and is there enough substance in your texts to be cited?

You need both, and it is the same project. A perfect answer on a website nobody knows is rarely cited. An authoritative company with vague, messy pages is read poorly. Think of it as a job interview: with AEO you give a clear answer, with GEO someone else vouches for you.

Where we stand: your buyer is already in the chat

Your buyer is already there. Forrester reported in January 2026 that 94 percent of business buyers use AI in the buying process, up from 89 percent a year earlier. Generative AI or conversational search was chosen twice as often as the most important source of information as any other source, and supplier websites lag far behind. According to OpenAI, ChatGPT has had 900 million weekly users since February 2026. Google is joining in. Anyone who has read the answer often no longer clicks through, so being visible in the answer itself becomes more valuable.

And most companies are looking the other way. According to a 2025 Yext report cited by Loganix, only 22 percent of marketers track their own visibility in AI and 64 percent are unsure how to measure success. Whoever starts now is ahead of the rest of the room.

Six things you can do now

Start by measuring where you stand, then improve the pages your buyer needs first. Six steps, in this order:

1. Play the buyer. Make a list of twenty questions that members of a DMU in your market ask, from buyer to engineer, and put them to ChatGPT, Perplexity, Gemini and Google's AI mode from a private browser or a computer other than your work one. Note who gets mentioned, whether you are among them and which sources are cited. It can be confronting. Repeat it every month with the same questions.
2. Put the answer at the top. Put the question in the heading and the answer in the first two or three sentences, with the explanation below. Do this per product, application and common choice. This is the AEO part of the work.
3. Back up what you write. Add figures, sources and quotes from customers or your own engineers. In the GEO study, exactly these three changes gave the biggest improvement in visibility. A tolerance, a standard and a customer name give an AI something to cite.
4. Put your facts in text. Specifications, certifications, capacity, materials and tolerances belong on the page as plain text, preferably not only in a PDF or image. Add Organization and Product markup and check that your robots.txt or cookie banner does not lock out AI crawlers.
5. Let others write about you. An AI also bases its answer on what others say about you. Trade media, industry associations, supplier databases, customer cases and reviews all count, and a mention in a trade journal certainly belongs there.
6. Make your information consistent everywhere. Company name, locations, offering and description should say the same thing on your website, LinkedIn, Google Business Profile and in industry directories.

Also check your analytics for traffic coming in from chatgpt.com and perplexity.ai. That is a rough but encouraging first indication that it is working.

Where it is heading: your next reader is a machine

The direction is clear and the pace is uncertain.

For industrial brands this means two things. AI helps you get on the shortlist, and after that your reputation decides with the people the buyer consults. The foundation for GEO and AEO therefore stays the same: clear answers, verifiable facts and third party mentions. What changes is the reader of your pages, because that is more and more often a machine. So write for that machine as if a person is looking over its shoulder.

Finally

Start this week with one hour. Put ten of those twenty questions to ChatGPT, Perplexity and Google's AI mode and write down which companies are mentioned. You then have a baseline and a list of pages to improve. That is all you need to get started.

Whoever is in the answer makes the shortlist. We make sure your brand gets chosen.

Want to know whether your company is mentioned when the DMU types in its question? We will go through your twenty questions together, in half an hour. Can we help you with your marketing challenge? A short introduction always delivers something valuable.