Loading...
E-commerce is changing faster than most shop operators want to admit. Three years ago, Google Shopping was the gold standard. Today consumers ask ChatGPT 'What is the best espresso machine under €300?' - and expect a direct recommendation, not a list of ten links. If you do not appear in that answer, you are not considered. This guide describes seven concrete steps with which online shops can systematically improve their ChatGPT visibility.
Step 1: Implement Schema.org Product markup completely. Schema.org is the universal standard for machine-readable product data - and the most important technical lever for AI visibility. A complete Product schema contains name, description, price, currency, availability, product image, aggregate rating, and product category. Critically: many shops implement only the mandatory fields. AI systems, however, also evaluate optional fields - in particular Brand, GTIN/MPN (unique product identifiers), delivery time, and return conditions. The more complete the markup, the higher the chance of appearing in AI-generated comparisons. Step 2: Provide an AI discovery feed. An llms.txt file in the root directory of your domain gives AI systems a structured overview of your offer. In addition, a machine-readable product feed in JSON-LD format is recommended. These feeds address multi-agent systems and future AI crawlers that explicitly look for machine-readable data. The effort is manageable; a one-time setup with automatic updates from the product catalogue is sufficient. Step 3: Allow GPTBot and relevant AI crawlers in robots.txt. Anyone who wants to appear in ChatGPT answers must not block the OpenAI crawler GPTBot. This sounds obvious, but in practice is surprisingly often the problem: many shops have inherited robots.txt rules that were originally intended for price-comparison portals and inadvertently also exclude AI crawlers. Check your robots.txt explicitly for GPTBot, OAI-SearchBot, ClaudeBot, and PerplexityBot.
Try it now
Check your GEO Score in 60 seconds - free, no account needed. 42 factors analyzed.
Step 4: Enrich product data substantively. Schema.org markup is the structure - but the content decides whether AI systems interpret your products. Concretely this means: product descriptions that explicitly name use cases and target audiences (not just listing features), FAQ schema beneath product pages with typical purchase-decision questions ('Who is this product suitable for?', 'What distinguishes it from model X?'), and Review schema with authentic ratings. AI systems such as Perplexity and ChatGPT with browsing capability actively read and weight customer reviews. A product with 200 reviewed ratings and an average of 4.3 is recommended more often than an equivalent product without reviews. Step 5: Set citation triggers. AI systems cite sources when content contains verifiable facts. For online shops this means: name concrete figures in product descriptions (battery life in hours, load capacity in kilograms, energy efficiency class), link to manufacturer specifications and test reports, and create your own content with measured values or comparisons you have conducted yourself. Example: instead of 'exceptionally long-lasting battery', write 'Battery life: 18 hours in real-world testing'. The second is citable - the first is not.
Step 6: Track mention rate and AI visibility. What is not measured cannot be improved. For e-commerce shops a two-track monitoring approach is recommended: first, manual spot checks - submit queries corresponding to your most important product categories in ChatGPT, Perplexity, and Gemini, and note when your brand appears. Second, automated monitoring via specialised tools that regularly send test sets to AI platforms and track the mention rate. Beconova offers this monitoring as part of the platform - nine AI engines are queried daily or weekly (depending on plan) with your key questions. Important metrics: Share-of-Voice (your share of mentions compared to competitors), hallucination rate (how often does an AI name wrong prices or product properties), and citation source (where does the information come from that leads to a mention?). Step 7: 4–6 weeks of iteration. AI visibility is not a one-off project. AI systems are regularly updated - training data, retrieval mechanisms, and ranking signals change. Plan fixed review cycles: every four to six weeks you check which steps you have implemented, which metrics have improved, and where new gaps have emerged. The majority of shops that start at all with AI visibility stop after the initial setup. Those who iterate win in the long run.
ChatGPT and Perplexity do not recommend products by chance. They draw on structured data, trustworthy sources, and consistent information. Online shops that invest now - implementing Schema.org markup completely, allowing AI crawlers, enriching product data, and tracking mention rate - build a lead that latecomers will find hard to close. Getting started costs time, not a large budget. And the effect can be measured.
Check GEO Score for freeMarvin Malessa
Founder, Beconova
Founded Beconova in Germany in 2025 to help shops and service businesses become visible in AI search engines. Writes about GEO, AI visibility, and the future of search.
Get started with Beconova now and optimize your presence in AI search engines.
Get Started