How AI recommends
products
When a customer asks ChatGPT "What is the best laptop for a university student?", the AI does not simply return a list of search results. It evaluates hundreds of sources, weighs expert reviews against consumer feedback, considers price-to-performance ratios and returns a curated recommendation of 2 to 3 specific products with explanations of why each one fits. This process happens billions of times per day across ChatGPT, Gemini, Perplexity and Google AI Overviews. Understanding how AI selects which products to recommend is the key to getting your products in front of the 30% of UK consumers who now research purchases through AI.
30%
of UK consumers research products via AI
2-3
products typically recommended per query
5x
higher conversion from AI recommendations
71%
of AI-cited pages use schema markup
How AI builds a product recommendation
When someone asks an AI platform for a product recommendation, the platform does not simply search a product database. It runs through a sophisticated process that combines multiple information sources, evaluates credibility and generates a recommendation that reads like advice from a knowledgeable friend. Understanding this process reveals exactly what your product pages and content need to contain to be selected.
Step 1: Intent classification
AI first determines what the user actually wants. "Best wireless earbuds" is a broad comparison request. "Best wireless earbuds for running under £80" is specific with a budget constraint. "AirPods Pro 2 vs Sony WF-1000XM5" is a head-to-head comparison. Each intent type triggers a different response format and different source priorities. For broad queries, AI draws heavily from roundup articles and buyer guides. For specific queries, it prioritises product pages with detailed specifications. For comparisons, it looks for content that evaluates both products side by side. Read more about how AI classifies queries in our guide on how AI understands search intent.
Step 2: Source gathering
AI gathers information from multiple source types. For product recommendations, the most influential sources are expert review sites (Which?, TechRadar, Wirecutter), retailer product pages with detailed descriptions, consumer review aggregators (Trustpilot, Amazon reviews), buyer guides from specialist retailers and manufacturer specification pages. ChatGPT combines training data with live Bing search. Gemini draws from Google Shopping data and Google Reviews. Perplexity searches the web in real time. Each platform weights sources differently, which is why the same question can produce different product recommendations across platforms.
Step 3: Credibility evaluation
Not all sources carry equal weight. AI applies credibility filters similar to E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness). A product review from Which? carries more weight than an anonymous blog post. A retailer with 2,000 Trustpilot reviews and a 4.7 rating is more credible than one with no reviews. A product page with detailed specifications, professional photography and genuine customer reviews is more trustworthy than one with minimal information. AI also checks for consistency: if multiple independent sources recommend the same product for the same use case, that product is more likely to be included in the recommendation.
Step 4: Recommendation generation
Finally, AI generates its response. For product queries, this typically includes 2 to 3 recommended products with brief explanations of why each is recommended, who it is best suited for and its key strengths and weaknesses. AI aims to provide balanced recommendations for different needs and budgets. "If budget is your priority, the JBL Tune 770NC offers excellent value at £79. For the best noise cancelling, the Sony WF-1000XM5 at £199 is the benchmark." This format means each recommendation slot is extremely valuable, and only products with strong, well-documented signals will be selected.
AI does not rank 100 products. It picks 2-3 and explains why. Getting into that shortlist requires strong signals across reviews, content and structured data.
Six factors that determine whether AI recommends your product
1. Review consensus
The single most influential factor for product recommendations is the consensus across reviews. AI does not just look at star ratings. It analyses review content to identify themes. If 200 reviewers consistently mention "excellent battery life" and "comfortable fit", AI will cite those specific attributes in its recommendation. Products with many detailed reviews that consistently praise specific features have the strongest recommendation signals. Encourage customers to mention specific features they liked in their reviews, not just give a star rating.
2. Expert validation
Products reviewed by established expert sources (Which?, TechRadar, Expert Reviews, Wirecutter, Good Housekeeping Institute) carry more weight in AI recommendations. If your product has been reviewed by a recognised expert source, make sure that review is easily findable online. Press coverage, awards (Which? Best Buy, Red Dot Design Award) and expert endorsements all strengthen the expert validation signal. For retailers, linking to or referencing expert reviews on your product pages helps AI connect your product to its expert validation.
3. Specification clarity
AI needs concrete specifications to make informed recommendations. A product page that states "16GB RAM, 512GB SSD, 14-inch 2560x1600 display, 72Wh battery (up to 12 hours), 1.4kg" gives AI everything it needs to recommend the product for specific use cases. A page that says "powerful performance and all-day battery life" gives AI nothing specific to work with. Include detailed specifications in both your page content and your Product schema markup. Specifications should be factual and measurable, not marketing language.
4. Price and value positioning
AI frequently structures product recommendations by price tier: a budget option, a mid-range option and a premium option. Your product's price positioning determines which slot it competes for. Make your pricing transparent and include it in your Product schema. AI also factors in value perception: a product priced at £149 that reviewers consistently describe as "outstanding value" may be recommended over a £99 product with mixed reviews. In the UK market, include VAT-inclusive pricing and mention delivery costs or free delivery thresholds prominently.
5. Product schema markup
Product schema tells AI exactly what your product is in a machine-readable format: name, brand, price, availability, aggregate rating, review count and detailed specifications. Without schema, AI has to parse your page HTML to extract this information and may miss key details. With schema, the information is structured and unambiguous. 71% of pages cited by ChatGPT use schema markup. For e-commerce, Product and Offer schema with AggregateRating are essential. See our guide on why structured data matters for AI.
6. Content depth and freshness
AI prioritises content that is both comprehensive and current. A buyer guide from 2024 is less likely to be cited in 2026 than a recently updated one. For product pages, include the release year, update dates and current pricing. For buyer guides and comparison articles, include the current year in the title and update regularly with new products and prices. Freshness is a strong signal for AI, particularly for product categories where new models launch frequently (electronics, fashion, home appliances). Stale content signals outdated information that AI will avoid citing.
Does AI recommend your products?
VestVale monitors whether ChatGPT, Gemini, Claude and Google AI mention your products. Track recommendations, competitor visibility and trends over time.
How each AI platform handles product recommendations differently
ChatGPT: review-driven recommendations
ChatGPT draws heavily from its training data, which includes product review sites, consumer forums and retailer content. It also searches the web in real time via Bing for current pricing and availability. ChatGPT's product recommendations tend to be conversational and explanatory: "The Sony WH-1000XM5 is widely considered the best noise-cancelling headphone. Reviewers consistently praise the comfort and sound quality. If you prefer a tighter fit for running, the Bose QuietComfort Ultra might suit you better." To influence ChatGPT recommendations, focus on getting your products reviewed by established review sites and maintaining comprehensive, up-to-date product pages with detailed specifications and genuine customer reviews.
Gemini: shopping data integration
Gemini has a unique advantage for product recommendations because it integrates directly with Google Shopping, Google Merchant Centre and Google Reviews. If your products are listed in Google Merchant Centre with complete data (price, availability, images, specifications), Gemini can access this structured information directly. Google AI Overviews, which are powered by Gemini, increasingly show product cards with images, prices and ratings for shopping queries. For UK retailers, having a well-maintained Google Merchant Centre feed is the single most impactful action for Gemini product visibility.
Perplexity: source-linked recommendations
Perplexity is the most transparent AI platform for product recommendations. It always cites its sources with clickable links, meaning each recommendation can drive traffic directly to your product page or buyer guide. Perplexity searches the web in real time and prioritises recent content. A buyer guide published this month will be cited over one from last year. For UK retailers, this makes Perplexity particularly valuable because it can generate direct referral traffic, not just brand mentions.
Claude: cautious and balanced
Claude tends to be more cautious with product recommendations than other platforms. It often presents multiple options without a strong preference, focusing on helping the user understand the trade-offs. Claude values authoritative, balanced content. A product page or guide that honestly discusses limitations alongside strengths is more likely to be cited by Claude than one that presents products as perfect in every way. For premium and B2B products, Claude is increasingly important as it is used by professionals for purchasing research.
Why products get different recommendations across platforms
Ask the same product question on four AI platforms and you will often get four different recommendations. This is because each platform draws from different sources, applies different credibility filters and generates responses differently. A product that dominates on ChatGPT (because of strong review site coverage) might be absent on Gemini (because the Google Merchant Centre listing is incomplete). To maximise your product's AI visibility, you need presence across all the source types that different platforms rely on. Our article about why AI results differ between platforms explains these differences in detail.
Which product categories are most affected by AI recommendations
Consumer electronics
Most affected category. Laptops, headphones, smartphones and smart home devices are among the most frequently asked product queries on ChatGPT. Detailed specifications and expert reviews are essential. Products reviewed by TechRadar, Which? or T3 have a significant advantage.
Home and garden
Growing rapidly. Mattresses, vacuums, garden furniture and home appliances generate high volumes of AI queries. These are considered purchases where consumers want expert guidance. Buyer guides comparing products by room size, budget and features perform extremely well.
Health and beauty
Significant and growing. Skincare routines, supplements and beauty devices are popular AI queries. Ingredient transparency, dermatological testing and genuine customer results are the strongest signals. Products with clinical backing or endorsements from qualified professionals have an edge.
Sports and fitness
Strongly affected. Running shoes, fitness trackers and exercise equipment are heavily researched through AI. Performance specifications, user experience reviews and suitability for different fitness levels drive AI recommendations. Specialist retailers with expert guides outperform generalists.
Baby and children
High-trust category. New parents rely heavily on recommendations for pushchairs, car seats and feeding equipment. Safety certifications, expert testing (Which? Best Buy) and detailed comparison guides are critical. AI is cautious with safety-related products and prioritises authoritative sources.
Fashion and clothing
Emerging impact. AI is increasingly used for fashion queries like "best waterproof jacket UK" or "sustainable fashion brands". Product descriptions emphasising materials, sizing accuracy and sustainability credentials perform well. Customer photos and detailed size guides strengthen AI visibility.
Frequently asked questions
Can I pay to have my product recommended by AI?
No. AI product recommendations are entirely organic. You cannot buy placement in ChatGPT or Gemini recommendations. The only way to influence recommendations is by building the signals AI looks for: strong reviews, detailed product information, expert validation and comprehensive schema markup.
Does Amazon dominate AI product recommendations?
Amazon is frequently mentioned but does not dominate the way it does in Google Shopping. AI recommends specific products, not retailers. A specialist shop with better product guides and more detailed reviews can be recommended alongside or ahead of Amazon for specific queries. Niche expertise beats general availability in AI recommendations.
How important is product photography for AI?
AI currently recommends products primarily based on text information, not images. However, Google AI Overviews are starting to show product images alongside recommendations. High-quality photography with proper alt text and image schema helps AI understand your products and may influence visual recommendation formats that are becoming more common.
Do product reviews on my own site matter for AI?
Yes. On-site reviews with proper Review schema are a valuable signal for AI. They complement third-party reviews (Trustpilot, Google) and provide product-specific feedback that AI can use in its recommendations. Genuine, detailed customer reviews on your product pages strengthen both your AI visibility and your credibility.
How does AI handle out-of-stock products?
AI platforms that search in real time (Perplexity, ChatGPT with web search) can detect stock status if your Product schema includes availability information. Products marked as out of stock may still be mentioned but with a note about availability. Keeping your stock status current in both your schema markup and Google Merchant Centre feed is important for accurate AI recommendations.
Should I create separate content for AI visibility?
You do not need separate content for AI. The same content that works well for AI also works for Google and for your customers. Detailed product pages, comprehensive buyer guides and genuine reviews benefit all channels. The key addition is structured data (schema markup) which is invisible to human visitors but essential for AI readability. Focus on depth, specificity and freshness.
Get your products recommended by AI
VestVale monitors whether ChatGPT, Gemini, Claude and Google AI recommend your products. Track visibility, competitor mentions and trends.
From £19.95/mo excl. VAT. Cancel monthly.