In six years of Oase Creative, few channel shifts have moved as fast as this one. Traffic from AI tools to US retail sites grew 138% year over year in May 2026, and according to Adobe Analytics those visitors converted 54% better than visitors from other sources. If you're not in the AI answer, you lose clicks: where Google shows an AI Overview, the clickthrough rate of the number 1 position in December 2025 was 58% lower than would have been expected without the AI Overview.
These aren't forecasts. This is what's measurable right now.
And yet the basics are still missing at many webshops. When we measured seventeen Dutch Shopify stores in August 2026, seven didn't even have a usable meta description on their category page (the measurement). Which is exactly where the difference gets made: not on product pages, and not on the homepage.
This article explains what Generative Engine Optimization is, why it works differently from SEO, and which seven elements decide whether AI cites you or your competitor.
What GEO is and why it's fundamentally different from SEO
SEO is about position. Rank 1, 2, or 3 in the organic results. You measure success in clicks, sessions, and rankings.
GEO has a different objective. Generative Engine Optimization is about whether you're included as a source when an AI system formulates an answer. Not clicks, but mentions. Not a position in a list, but inclusion in the answer itself.
The mechanics differ too. Google's classic algorithm weighs backlinks, keyword density, and on-page signals. AI systems like ChatGPT, Perplexity, and Google's AI Overviews weigh something else: structure, semantic consistency, verifiable data, and the degree to which your content contains self-contained information units that are useful without further context.
One number makes the difference visible. Ahrefs compared 300,000 keywords in December 2023 and December 2025: for keywords that now show an AI Overview, the clickthrough rate of the number 1 position fell 58% further than would have been expected from comparable keywords without one. The click goes to the source cited in the AI answer, or nowhere.
GEO doesn't replace SEO. A strong SEO foundation (domain authority, a technically sound site, relevant backlinks) actually increases your chance of AI citation. But SEO alone is no longer enough. For the broader context of AI search in e-commerce, see GEO for webshops.
The question is no longer only: does my page rank? The question is also: does my page get cited?
Why category pages are the key, not product pages
Category pages matter more for GEO than product pages, because AI systems answer questions and category pages cover those questions. Yet most GEO advice is about product pages. Better titles, more reviews, stronger bullet points. All useful. But most of your GEO attention belongs on category pages, not on individual SKUs.
The reason lies in the nature of AI citation. AI systems answer questions. What's the best vacuum cleaner for an apartment? Which coffee machine suits me? What's the difference between a stick blender and a regular blender? Pages that answer those questions exhaustively position best for citation.
A product page for one specific Miele vacuum model can't do that. A category page about vacuum cleaners for small spaces, with comparisons, FAQs, and guides, can.
AI systems structurally prefer pages that:
- Make comparisons between options
- Answer questions from a broad perspective
- Contain transparent and verifiable data
- Are updated regularly
That profile fits a well-built category page far better than a product page. And yet in most webshops, category pages are still thin pages with a product grid and some boilerplate text underneath.
That's the opportunity.
Element 1: FAQ sections with structured data
The most efficient GEO investment per word is an FAQ section with correctly implemented FAQPage schema.
Three effects at once. An FAQ quickly adds a few hundred up to roughly a thousand words to your category page with minimal visual footprint: the questions and answers sit compactly. They cover long-tail search terms you can't fit into regular page text. And they hand AI systems ready-made information units: a question plus a self-contained answer. Note: Google removed the FAQ rich result from the search results entirely on 7 May 2026 (Google Search Central). So the gain here sits in AI citation and content density, not in expandable questions within Google itself.
That last point is critical. ChatGPT and Perplexity prefer to cite content you can lift verbatim into an AI answer without further context. An FAQ answer meets that condition.
Implementation requires two things. One: the FAQ section exists as HTML on the page, not only loaded dynamically by JavaScript, because AI crawlers don't always see dynamic content. Two: each question and answer is marked up with the FAQPage schema in JSON-LD. Google no longer does anything visible with it in the search results, but it remains a valid schema.org type that makes the question-and-answer structure unambiguous for any system that reads the page.
Write questions customers actually ask. Not the questions you find convenient. Check your customer service inbox. Look at the "People also ask" section in Google. Those are the questions AI answers want to resolve.
For the structural setup of category pages, see our product listing design service: the whole framework is built into that work.
Elements 2 and 3: Comparison guides and transparent product data
AI systems cite sources that make comparisons. Variant A vs. variant B is one of the strongest content formats for GEO, precisely because it follows the pattern of the questions users ask AI assistants.
A category page for outdoor blinds, for example, gets a section comparing cassette blinds with freestanding models. Not a list of products, but a written-out comparison: when do you choose one, when the other, what are the criteria?
Comparison guides work for a specific reason: they're informational but not direct sales copy. AI systems avoid citing content that's overtly commercial. A comparison is neutral enough to be cited, yet it lives on your domain and brings the visitor to your products.
Transparent product data is the second element in this group. Price, delivery time, warranty terms, material specifications. When that data sits on your page, accurate and consistent, it raises your chance of AI citation. Inconsistencies do the opposite: AI systems avoid sources where data conflicts or stays unclear.
That sounds simple. In practice we see at Oase Creative that many category pages have friction here: delivery times that don't match what's actually delivered, warranty periods that appear on product pages but not in the category description, no price indication at category level. Every inconsistency is a reason for AI to cite a different source.
Elements 4 and 5: Review schema and external mentions
Reviews are relevant in two ways. Substantively, customer reviews are proof that a product works: AI systems value externally verified signals. Technically, reviews with the right schema markup are directly readable by AI crawlers as structured data.
Review schema (the AggregateRating type in JSON-LD) tells AI systems: this page has X ratings averaging Y. That's a verifiable signal of quality and popularity. Product pages are the most logical place for review schema, but on category pages you can also display and mark up summarized rating data.
External mentions work differently but matter just as much. When your brand, your products, or your category pages get mentioned in media articles, comparison sites, blogs, or partner content, AI systems factor that in as a trust signal. It's comparable to backlinks for SEO, but for AI citation: external sources that reference you raise the chance that AI treats you as a reliable source.
Concrete actions for external mentions: press releases on product updates, contributions to trade publications in your sector, partnerships with creators in your niche who link to your category pages, and participation in comparison platforms that editorially include your products.
The combination works cumulatively. A category page with review schema plus external mentions pointing to it is structurally stronger than a page that leans on on-page signals alone.
Elements 6 and 7: Internal link structure and content freshness
Internal linking is already an underrated factor in classic SEO. For GEO it's underrated just as much.
AI systems crawl your site like Google does: they follow links. A strong internal link structure from your category pages to relevant blog articles, comparison guides, and service pages helps AI understand the coherence of your content. It positions your domain as an authority in a topic area, not as a loose collection of pages. For a related pitfall on the content side, see how AI-generated content costs you rankings.
In practice: a category page about standing desks links to a blog article on workplace ergonomics, to delivery information, and back to your about page if relevant expertise is described there. Reciprocal links strengthen the thematic cluster.
For product photography, the same principle applies: the more relevant internal pages point to that cluster node from quality context, the stronger the thematic signal.
Content freshness is the seventh element. AI systems prefer content that has been recently updated. That doesn't mean rewriting every category page monthly, but it does mean having a deliberate update strategy. Add new comparison data when products change. Update FAQ answers when customer questions shift. Refresh statistics when new data becomes available.
A simple way to signal freshness: put a "Last updated" date at the top of your category page and keep it current. That's a minimal gesture that sends a visible signal to both human visitors and AI crawlers. Only change the date when something has actually changed: a date that moves without new content is not an update.
The seven elements at a glance
| Element | What it gives AI systems | Quickest first step |
|---|---|---|
| 1. FAQ section with FAQPage schema | Ready-made question-and-answer pairs | Answer five real customer questions from your inbox |
| 2. Comparison guide | The answer to "X or Y?" | One written-out comparison on your best-selling category |
| 3. Transparent product data | Prices, delivery times and warranty that match everywhere | Put category and product pages side by side and look for contradictions |
| 4. Review schema | A verifiable quality signal | AggregateRating in JSON-LD on pages with reviews |
| 5. External mentions | Confirmation from outside your own domain | Approach one trade publication or comparison platform in your niche |
| 6. Internal links | Coherence between category, guide and explanation | Link from the category page to your buying guide and back |
| 7. Freshness | A reason to pick your version over an older one | A fixed moment each quarter to review figures and FAQs |
New KPIs: how to measure GEO success
Classic dashboard metrics haven't disappeared. They're just no longer enough.
Organic traffic and rankings tell you what's happening with your classic SEO. For GEO you need additional KPIs that directly measure whether you're being cited.
AI Overview visibility. Since 3 June 2026 Search Console has dedicated reports for generative AI, rolled out to all sites worldwide on 31 August 2026. You see per page how often your URLs appeared in AI Overviews, AI Mode and Discover, by country, device and date. Clicks are not included yet. Supplement it with manual work: search weekly on 5 to 10 informational queries relevant to your category pages and note which source is cited.
Brand mentions in AI answers. Ask ChatGPT, Perplexity, and Gemini about products in your category. What are the best [category] for [use case]? Is your brand or webshop named? That's your baseline. Run it now, before you make GEO improvements. Repeat after three months.
Branded search growth. When AI systems cite you, more people then search directly for your brand. Branded search growth is an indirect but reliable indicator of the AI citation effect.
AI traffic in analytics. Track in GA4 how many visitors arrive via chatgpt.com, perplexity.ai, or other AI sources. In Adobe's May 2026 measurement, that visitor group converted 54% better than visitors from other sources. Do you know how large it is for you? Part of it arrives as direct traffic; how to make that visible is covered in measuring AI referrals in GA4. Our service page on SEO covers how we wire this into reporting.
| KPI | Measurement source | How often | Baseline now |
|---|---|---|---|
| Impressions in AI Overviews and AI Mode | Search Console, generative AI report | Monthly | Note impressions per category page |
| Brand mentions in AI answers | Fixed set of questions in ChatGPT, Perplexity, Gemini | Quarterly | Save the answers with a date |
| Traffic from AI sources | GA4, referral source | Monthly | Sessions and conversions per source |
| Branded search | Search Console, performance report | Monthly | Clicks on your brand name |
What this means for your roadmap
Two paths, depending on where you stand.
If you don't yet have FAQ sections with FAQPage schema on your category pages: that's element 1, and it's by far the quickest win. Count on half a day of work per category to write the questions, plus a development hour to implement the schema.
If FAQs are already in place: move to elements 2 and 3. Build comparison guides on your three best-converting category pages and audit the consistency of delivery times, warranty terms, and prices between your category and product pages.
The webshops investing in this now are building a sales channel that becomes substantially larger in a few years. The rest wait until it's too late.
Want us to look at this together? Book a call. We'll walk through your three toughest category pages and give you a concrete priority list.
