page index
Share this article:
A page can sit at position 3.6 on Google and still bring not a single click. And a company can rank well on Google and still fail to appear in AI answers. This article explains how SEO, AEO, and GEO differ, what we measured for lean-sharp.ch across 27 queries in four AI systems, and where you should start.
What is the difference between SEO, AEO and GEO?
SEO ranks your pages in classic search results; AEO (answer engine optimization) makes your answer appear directly, for example in a snippet or an AI overview, and GEO (generative engine optimization) gets AI systems such as ChatGPT to name your company in their answers. For Google Search, AEO and GEO remain SEO.
Quick Facts:
- Our measurement: Across 27 queries in ChatGPT, Claude, Gemini and Perplexity, lean-sharp.ch was named 9 times, in 5 of 8 growth hacking answers and in 4 of 19 answers on SEO, GEO, and AEO.
- Ranking is not impact: For “aeo agentur in zürich” we sit at position 3.6 in Search Console, yet receive 0 clicks.
- Measuring: Besides rankings, brand mentions and citations in AI answers count, checked platform by platform.
SEO, AEO and GEO at a glance
The three terms do not describe separate disciplines. They describe three places where your customers find you today: in the list of results, in the direct answer, and in the text of an AI.
| Dimension | SEO | AEO | GEO |
|---|---|---|---|
| Goal | A good position in the list of results | Your answer is shown directly | Your company is named in AI answers |
| Where it appears | Google, Bing | Snippets, AI overviews, voice assistants | ChatGPT, Perplexity, Claude, Gemini, AI Mode |
| Typical metric | Position, clicks, enquiries | Snippet share, impressions without a click | Brand mention, citation, share of AI answers |
| Typical lever | Technology, content, internal links, backlinks | A clear answer under a question heading | Verifiable content, mentions outside your own website |
| Risk if ignored | Your page is not found | The answer appears for others | You are missing from the recommendation although you rank |
The table summarizes the differences as an assessment by Lean & Sharp. It contains no measured values.
From practice: what we measured for lean-sharp.ch
Re-measurement on 25 October 2026: 27 queries across four AI systems
We asked fully phrased search queries live in ChatGPT, Claude, Gemini, and Perplexity. They come from Search Console and from a database of real AI queries, 27 queries in total. All queries were in German and aimed at the Swiss market. For Google AI Overview, we checked three SEO queries from that database.
| Platform | Queries | Lean & Sharp named | With own source on |
|---|---|---|---|
| ChatGPT | 9 | 3 | not measurable in the test |
| Claude | 6 | 2 | 2 |
| Gemini | 6 | 3 | 3 |
| Perplexity | 6 | 1 | 0 |
For growth hacking queries we were named in 5 of 8 answers, for SEO, GEO and AEO queries, in 4 of 19. In Google AI Overview we did not appear for any of the three queries. Each of the two competitors we checked was named in 3 of the 27 queries.
Two patterns stand out. First, Claude and Gemini cite our English pages, for example the English SEO service page. The German service page was not cited in any answer. Second, we lose out on queries without a location: for “beste seo agenturen” (best SEO agencies), Claude, Gemini, and Perplexity mostly named German agencies and ChatGPT named Swiss providers. We did not appear in any of the four systems.
Keywords are not prompts
An earlier run with 14 short keywords such as “seo agentur” produced not a single mention. But short search terms make poor prompts because AI systems answer them in a vague or generic way. With fully phrased questions, the picture looks different. The two runs are therefore not directly comparable.
Perplexity follow-up questions are not search queries
During the test, Perplexity showed follow-up questions, with “I want to compare Lean & Sharp as a market overview” among them (in German in the original). We considered adding such suggestions to the prompt list and decided against it. They come from the AI and not from a person who typed them, and they have no search volume. As a hint about missing content formats they are still valuable: the suggestions “comparison dashboard” and “checklist of warning signs” show what may be missing for this question.
Search Console: position 3.6 for “aeo agentur in zürich”, but 0 clicks
For the last three months (queried on 25 October 2026), Search Console shows the same pattern from the other direction. For “aeo agentur in zürich” (“AEO agency in Zurich”) we are on average at position 3.6 with 116 impressions, and there were 0 clicks. For “aeo agentur schweiz” (“AEO agency Switzerland”) we are at position 15.5 with 169 impressions.
It is also striking that fully phrased questions appear, for example “welche geo (generative engine optimization) agentur gibt es in zürich?” (“which GEO agencies are there in Zurich?”) with 25 impressions at position 5.9. AI systems answer exactly this kind of question with a list of recommendations, not with a list of results.
Our assessment: a ranking in the top 5 without clicks may be because the title and description do not match the question or because the answer is already shown on the results page. The data alone cannot separate the two. All we know is this: ranking and impact are two different measures, and you have to measure the second one separately.
What the measurement does not show
Part of the queries come from our own Search Console, that is, from questions for which we already had impressions. This skews the result in our favour. ChatGPT returned no source information in the test, so only mentions count there. The sample is small and a snapshot in time.
Where you should start
Which layer takes priority depends on the symptom. The table is our assessment from projects, not a measurement series.
| Symptom | Likely cause | First step |
|---|---|---|
| Page ranks in the top 10 but brings few clicks | Title and description do not match the search question or the answer is already on the results page | Align title, description and the first sentences with the concrete question |
| Important topic, page does not rank | Technology, structure, or content are not sufficient | Check the SEO basics: indexing, page structure, internal links |
| Page ranks well, your brand is missing from ChatGPT answers | Your brand is hardly mentioned outside your own website | Build mentions, for example with expert posts, see LinkedIn as an SEO Channel |
| Competitors are named in AI answers, you are not | Your competition is better documented or mentioned more often | Competitor comparison with real search queries, platform by platform |
How to structure a single page so that it works as an answer is covered in our article on AEO-optimized content structure. This article is about which layer comes first for you.
What Google itself says
In July 2026, Google published its own guide to optimizing for generative AI features. For Google Search this means:
- AEO and GEO are still SEO. The AI features draw on the same quality and ranking systems.
- Unique first-hand content is particularly effective, not summaries of general knowledge.
- Not necessary are llms.txt, splitting content into small snippets, or building a separate page for every search variant.
- Structured data is not required for Google’s AI features but remains useful for rich results.
The guide applies to Google Search. ChatGPT, Perplexity, Claude, and Gemini have their own retrieval and citation mechanisms, which is why we check each platform separately instead of speaking of “the AI”.
How to measure AEO and GEO
You still track rankings and clicks in Search Console and GA4. For the AI layer, a prompt test that you can run yourself in five steps helps:
| Step | What you do | What to watch for |
|---|---|---|
| 1. Collect queries | Take search queries from Search Console, from sales conversations and from “People also ask” on Google | Use real wording, not invented wording |
| 2. Define platforms | Query ChatGPT, Google AI Overview, Perplexity, Claude, and Gemini one by one | Not every query triggers an AI overview on Google, note this separately |
| 3. Evaluate answers | Record brand mention, citation of your website, and named competitors separately | A mention in the text is not the same as a link to your page |
| 4. Compare competitors | Evaluate the same queries for two to three competitors | Shows whether a gap exists across the industry or only for you |
| 5. Repeat | Repeat the test at a fixed interval, for example once per quarter | Answers change, a single test is a snapshot |
The table describes our approach and is not a measurement series.
Three metrics follow from this:
- Brand mention: In how many of the checked search queries does the platform name your company?
- Citation: In how many answers does your website appear as a source and which page?
- Competitor share: How often are two to three competitors named for the same queries?
According to Google, Search Console also offers a performance report for generative AI features. Figures from third parties are an additional signal, not internal Google truth. How quickly measures take effect depends on the topic and the competition. We do not promise fixed timeframes.
Conclusion
SEO, AEO, and GEO are not alternatives. They are three measuring points for the same question: do your customers find you where they search today? Our own measurement shows that a good ranking does not answer this question.
The most sensible first step is therefore a measurement rather than a measure. If you know for which search queries you are named and for which you are not, you can decide deliberately whether to work on placement, on the answer, or on brand presence.
Do you know for how many search queries your company appears in AI answers today?
In a free, no-obligation conversation, as short as a cup of coffee, we look together at where you stand on SEO, AEO, and GEO.
Share this article:
Share this article:
Are you dissatisfied with the results of your digital campaigns for search engine optimization (SEO), paid advertising, content marketing, or gamification? Let’s start with a short call for a free and non-binding exchange about growth hacking and digital marketing.
Dr. Gero Kühne
Owner
FAQs
No. For Google Search, GEO remains part of SEO, and the other AI systems also rely on indexed, well-maintained websites. Without solid SEO foundations, it is hard to be named in AI answers.
With a measurement of the current situation: how do your most important pages rank, and is your company named in AI answers for real search queries? After that you can decide whether technology, content, or brand presence comes first.
Not for Google’s AI features, according to Google’s own guide. Structured data remains useful for rich results. For other platforms, we would not present schema markup as a prerequisite, but as a supplement.
That depends on the topic, the competition, and the platform, and we cannot promise fixed timeframes. A prompt test at a fixed interval, for example once per quarter, shows whether anything is moving.
We are a boutique agency for growth hacking and digital marketing and work on the principle of practice before theory. That means our own tests and data, for example GEO snapshots and self-audits of our own website, instead of hype.
