Case study: how a webshop tripled AI citations
A webshop that AI could not find
The webshop in this case study sells sustainable household products, from reusable kitchen items to eco-friendly cleaning products. With over 2,000 products, an active blog and a growing customer base, the website was well positioned in traditional search engines. But when asking ChatGPT or Perplexity about sustainable product recommendations, the webshop appeared nowhere.
The owner discovered the problem when a customer mentioned she had found the webshop via Google, not via Perplexity where she normally conducted her product research. Upon inquiry, it turned out that Perplexity consistently recommended three competitors but never mentioned this webshop.
This scenario is recognizable for many e-commerce entrepreneurs. As we describe in our article about AEO and its impact on your website, product research is increasingly moving to AI-powered platforms. A webshop that is not cited in AI answers misses a growing segment of customers.
The webshop has approximately 2,000 products, 150 blog articles and a monthly visitor count of 45,000. The team consists of the owner, a content creator and a part-time developer. Names have been anonymized for privacy reasons.
The diagnosis: what AI models were missing
A thorough analysis of the website revealed five core problems that hindered AI visibility.
- Product pages had no Product schema markup. AI models could not read product information in a structured way.
- Category pages consisted purely of product listings without descriptive content that provides AI models with context.
- The blog covered topics superficially in short posts of 300 to 500 words, insufficient for AI citation.
- There were no FAQ sections on product pages, while customers asked the same questions daily through customer service.
- The robots.txt allowed AI crawlers, but the sitemap was outdated and contained hundreds of 404 pages.
The absence of Product schema proved to be the biggest gap. In our article about Schema.org markup, we explain why structured data is essential for e-commerce websites pursuing AI visibility. Without schema, you are not speaking the language of AI models.
Phase 1: transforming product pages with structured data
The first and most impactful step was implementing comprehensive Product schema markup on all product pages. The team used an automated script that extracted product data from the CMS and converted it to JSON-LD.
<script type="application/ld+json">\n{\n "@context": "https://schema.org",\n "@type": "Product",\n "name": "Reusable Beeswax Wraps (Set of 3)",\n "description": "Sustainable alternative to plastic cling wrap...",\n "brand": {\n "@type": "Brand",\n "name": "EcoWrap"\n },\n "offers": {\n "@type": "Offer",\n "price": "24.95",\n "priceCurrency": "EUR",\n "availability": "https://schema.org/InStock"\n },\n "aggregateRating": {\n "@type": "AggregateRating",\n "ratingValue": "4.7",\n "reviewCount": "128"\n },\n "review": [\n {\n "@type": "Review",\n "author": { "@type": "Person", "name": "Maria K." },\n "reviewRating": {\n "@type": "Rating",\n "ratingValue": "5"\n },\n "reviewBody": "Fantastic alternative, I use them daily."\n }\n ]\n}\n</script>The team implemented not only the basic fields but also added aggregateRating, reviews and detailed product specifications. The richer the schema markup, the better AI models can interpret and compare product information with competitors.
Phase 2: enriching category pages with content
Category pages in e-commerce are often little more than a grid of product images with prices. For AI models, they offer barely usable information in that form. The team transformed the twenty most important category pages by adding editorial content.
- Each category page received an introduction of 200 to 400 words describing the product type, explaining the benefits and offering tips for choosing the right product.
- Below the product listing, a FAQ section was added with five frequently asked questions about the product category, complete with FAQPage schema.
- Internal links to relevant blog articles were added as "Learn more about..." sections.
- A comparison table was included for categories with three or more comparable products.
This approach gave AI models the context they needed. Instead of just product data, they now also received buying advice, comparisons and background information they could cite in answers to consumer questions.
Phase 3: transforming the blog into an authority platform
The existing blog consisted of 150 short posts about seasonal topics. The team selected the thirty best-performing posts and rewrote them into in-depth articles of 1,500 to 2,500 words.
Each rewritten post followed a fixed structure: a clear introduction, logical H2 and H3 sections, practical examples, a summary with five key points and a FAQ section. This aligns with the principles we describe in our articles about heading hierarchy and readability for AI.
Additionally, twenty new articles were written around product-related questions that customers regularly asked. Topics such as "How do you maintain beeswax wraps?" and "Which sustainable cleaning product for which surface?" were developed into comprehensive guides. Each article contained links to the relevant product pages.
The combination of product pages with schema markup and blog articles with in-depth buying guides created a knowledge network that AI models evaluate as a whole. The blog proved the expertise, the product pages delivered the concrete data.
The results: from 8 to 27 citations per month
After eight weeks of optimization, the team measured results by weekly asking 50 relevant product questions to ChatGPT, Perplexity and Google Gemini.
AI citations per month (50 test questions, 3 platforms)
Before optimization: 8 citations (5.3% citation rate)
After 4 weeks: 16 citations (10.7% citation rate)
After 8 weeks: 27 citations (18.0% citation rate)
Distribution per platform (after 8 weeks):
Perplexity: 14 citations (most, due to real-time fetching)
ChatGPT: 8 citations
Google Gemini: 5 citations
Most cited content:
Buying guide blog articles: 12 citations
Product pages with reviews: 9 citations
Category page FAQs: 6 citationsImpact on revenue and traffic
- Organic traffic increased by 41% over the eight-week measurement period.
- Traffic from AI platforms (identified via referrer analysis) grew from 2% to 9% of total traffic.
- The conversion rate of AI-referred visitors was 23% higher than that of regular organic traffic.
- The average order value of AI-referred customers was 18% higher, suggesting these visitors are already better informed.
- The return rate decreased by 12%, possibly because customers made better product choices through AI answers.
Lessons for other webshops
Five concrete lessons emerge from this case study that every e-commerce entrepreneur can apply.
First: invest in Product schema markup as the top priority. Of all optimizations, this had the most direct impact on AI citations. AI models use structured product data to substantiate product recommendations.
Second: make category pages more than product listings. The editorial content on category pages proved to be an unexpectedly strong source of AI citations, especially the FAQ sections.
Third: write buying guides that compare product categories. This type of content perfectly matches how consumers use AI models for product research.
Fourth: use customer review data in your schema markup. AI models value social proof and are more likely to cite products with high ratings and many reviews. This relates to the E-E-A-T principles where "Experience" is demonstrated through real user experiences.
Fifth: monitor which questions consumers ask AI models about your product category. Those questions are the foundation of your content strategy. Every frequently asked question is an opportunity to be cited.
In e-commerce, AI visibility is no longer a luxury. It is the difference between being found at the moment a customer has purchase intent and being completely overlooked.
Dive deeper: Schema.org markup: the language AI understands | MCP servers: how AI agents communicate with your website | A2A protocol and agent communication
Key takeaways
- A webshop with 2,000 products tripled AI citations from 8 to 27 per month in eight weeks through targeted optimization.
- Product schema markup on all product pages was the single most impactful measure.
- Category pages enriched with editorial content and FAQ sections were cited by AI models surprisingly often.
- In-depth buying guides on the blog generated the most individual citations.
- AI-referred visitors convert better and order more, making the investment in AEO optimization directly profitable.
Frequently asked questions
Do all 2,000 product pages need schema markup?
Ideally yes, but you can prioritize. Start with your best-selling products and products for which consumers most frequently ask AI models for advice. An automated script that generates schema from your product database makes this scalable. Most e-commerce platforms offer plugins or extensions that automatically generate Product schema based on your product catalog fields.
How do I know which product questions consumers ask AI?
There are multiple methods. Ask questions to ChatGPT and Perplexity about your product categories yourself and analyze how they respond. Review which questions your customer service receives most frequently. Use Google's "People Also Ask" for product-related search terms. Analyze the search terms in your own site search engine. Each of these sources provides valuable input for your content strategy.
Does this approach also work for webshops with fewer products?
Absolutely. A smaller product catalog actually makes it easier to thoroughly optimize each product page. A webshop with 50 products can manually equip each page with comprehensive schema markup, detailed descriptions and FAQ sections. The principles are scalable in both directions.
How long until product pages get cited?
With Perplexity, which fetches pages in real-time, the effect can be visible within days once your pages are indexed. With ChatGPT it takes longer because the model is periodically updated with new training data. Expect two to six weeks for Perplexity, four to twelve weeks for ChatGPT and varying timeframes for Google Gemini, which combines both real-time sources and trained knowledge.
What is the effect of customer reviews on AI citations?
Customer reviews have a demonstrably positive effect on AI citations. AI models use review data as a signal of product quality and reliability. Products with more than 50 reviews and an average rating above 4.0 are cited significantly more often than products without reviews. Including review data in your Product schema makes this information directly available to AI models.
Every product page is an opportunity to be cited by AI. The webshops that invest in structured product data and in-depth buying guides will be the winners in the era of AI-powered shopping.
How does your website score on AI readiness?
Get your AEO score within 30 seconds and discover what you can improve.
