Generative AI has officially moved from searching to shopping. With the launch of ChatGPT Shopping Research, customers can now ask for product recommendations directly inside ChatGPT – from “best waterproof jacket for winter under £150” to “ideal football boots for amateur league play this season” and receive personalised guides built from web sources, product feeds, reviews, specs, and merchant data.
I tested ChatGPT’s new Shopping Research feature using a simple query – “best trail running shoes“ and was genuinely fascinated by how interactive and dynamic the experience was. The results weren’t just product suggestions; they included comparisons, reasoning, pros and cons, and follow-up options that felt much closer to speaking with an expert than using a traditional search engine. It’s a small glimpse into how AI will reshape product discovery – more conversational, more personalised, and far more intuitive than today’s typical eCommerce journey. The video below walks through exactly what I saw and why this shift matters for retailers preparing for AI-driven shopping.
For retailers, it changes the fundamental question from: How do we rank in Google? → How do we get retrieved by LLMs?
And the next phase goes even further: Execution. By 2026, AI agents won’t just retrieve products – they will act: compare retailers, add to basket, validate stock, navigate checkout, and complete purchases autonomously.
AI Shopping Strategy for Retail Brands to Improve Product Discovery
To make this guide easier to navigate, here is a quick links menu to jump directly to each major section:
- 1. Own Your Product Data – The New Source of Truth for LLM Retrieval
- 2. Maximise “Retrievability” in AI Shopping Assistants
- 3. Deepen Reviews & First-Party Sentiment Data
- 4. Optimise Merchant Feeds for AI Systems
- 5. Enable the Transaction: The Agentic Protocol (Agentic Commerce)
- 6. Strengthen Brand & Entity Signals
- 7. Prepare for the New KPI: AI Visibility
1. Own Your Product Data – The New Source of Truth for LLM Retrieval
AI shopping assistants depend on structured, machine-readable product data. Your product information is now the primary signal LLMs use to understand, compare, and recommend products.
AI shopping systems rely on:
- Product feeds (Google Merchant Center, Amazon, marketplace feeds, future OpenAI merchant integrations)
- Schema.org structured data
- Price, availability, fulfilment & delivery signals
- High-quality, attribute-rich PDP content
- Review depth, recency, and sentiment
To appear in AI-generated recommendations, product data must be consistent, complete, and unambiguous.
LLMs typically retrieve only a handful (3–6) products per query, meaning retailers are competing for an extremely limited number of AI recommendation slots. Missing or unclear attributes can exclude your products entirely.
Product data quality now functions as a core “ranking factor” for AI retrieval – incomplete, outdated, or ambiguous feeds are simply skipped, regardless of your traditional SEO strength.
Structured data must be server-rendered, not injected late via JavaScript. LLM scrapers often miss client-side JSON-LD.
Actions for Retail Brands
a) Standardise and strengthen structured product data.
As part of your internal SEO checklist, ensure all products include these core schema attributes:
namebrandmodelcolourcategoryofferspriceavailabilityaggregateRatingreviewimage
Then enrich this with deeper, retrieval-friendly attributes:
- Material
- Fit / cut
- Technology (e.g., cushioning, insulation, fabric innovations)
- Primary use-case (road running, office wear, travel, home gym)
- Sport or activity
- Closure type
- Collection or season
- Gender
These should be reviewed and updated regularly — think of this as your “PDP data quality checklist.”
b) Optimise for Computer Vision
LLMs are multimodal: they see your images and verify whether your descriptions match reality.
- Clean backgrounds improve object recognition
- Multiple angles showing each claimed attribute (stitching, sole patterns, buckles)
- Alt Text 2.0: Describe the visual scene accurately
c) The ‘Values’ Filter (Sustainability & Ethics)
- Add sustainability or eco attributes (recycled %, origin, certifications)
- Include repairability, durability, and longevity data
- Add energy/material efficiency details where relevant
2. Maximise “Retrievability” in AI Shopping Assistants
SEO is shifting from Ranking → Retrieval.
AI shopping tools (ChatGPT Shopping Research, AI Overviews, Perplexity, Gemini) retrieve products directly from structured data, feeds, and authoritative content.
Your goal as a retailer is to become the most machine-readable, machine-trustworthy source in every category.
Static, lightweight HTML is now essential. JavaScript-blocked or late-rendered components increase the risk that AI crawlers cannot access price, reviews, or key attributes – causing your products to be excluded from AI results.
Niche and specialist retailers with deep, attribute-rich product data are increasingly outperforming large generalists in AI retrieval. Depth beats breadth in LLM-driven commerce.
Actions for Retail Brands
a) Make your site architecture LLM-friendly.
Add these items to your technical SEO checklist:
- Use clean, semantic HTML
- Serve static HTML for core content (reduces token cost for AI models)
- Keep category paths simple and predictable
- Maintain consistent canonicalisation rules
- Ensure PLP and PDP content loads server-side or is pre-rendered
b) Create AI-optimised category guides.
AI assistants increasingly surface curated collections such as:
- Best running shoes under £100
- Warmest puffer jackets for winter
- Top gifts for travellers
Your category content checklist should include:
- Attribute-driven comparison tables
- “How to choose” guidance
- Buyer’s guides
- FAQs
- ItemList + FAQPage schema
- Internal links to top products
c) Build authoritative product comparison pages.
Consumers increasingly ask:
- “Which is better, X or Y?”
- “Is this worth the extra £50?”
- “What’s the best option for beginners?”
Ensure your content plan includes comparison pages for:
- Similar products from different brands
- Generations of the same product
- Budget vs premium options
- Best items for specific use cases
3. Deepen Reviews & First-Party Sentiment Data
AI models use sentiment signals to decide which products to recommend. Review volume, recency, and quality directly influence visibility in AI shopping answers.
ChatGPT frequently surfaces descriptors like “best budget pick” or “durable option” directly from aggregated review sentiment and third-party citations. These off-site signals matter more than ever.
Actions for Retail Brands
- Grow verified review volume (email, loyalty, incentives)
- Implement
ReviewandAggregateRatingschema - Highlight pros/cons or attribute ratings
- Ensure review data flows into product feeds
- Surface real user Q&A
This should be part of a recurring CRM + CRO checklist.
4. Optimise Merchant Feeds for AI Systems
Your product feed is becoming more important than your PDP. AI shopping systems rely heavily on real-time, structured product feeds—they are predictable and easier to parse than full web pages.
ChatGPT Shopping Research frequently mirrors Bing Shopping’s product data. Because Bing Merchant feeds feed into ChatGPT’s retrieval layer, optimising for Bing can increase your AI visibility.
Actions for Retail Brands
a) Upgrade product feeds from compliant → AI-optimised
Your feed optimisation checklist should include:
- Material
- Use case
- Technology
- Colour & size variants
- Gender & audience
- Care instructions
- High-quality images
- Alt text
- Model/series names
b) Ensure cross-system consistency
Align:
- Product feed titles
- On-site titles
- Structured data
- Breadcrumbs
- Category metadata
- Image alt text
- Naming conventions
5. Enable the Transaction: The Agentic Protocol (Agentic Commerce)
The next evolution of AI shopping is not just retrieval but execution. By 2026, AI agents will add items to carts, check stock, navigate checkout flows, and complete purchases autonomously.
These considerations should now be added to your “Agentic Commerce readiness checklist”:
Emerging agentic shopping flows show exceptionally high conversion rates (10–15%). Even though AI referral volume is small today, each inclusion is disproportionately valuable.
a) API-First Inventory & Pricing
- Live stock levels
- Real-time price
- Variant availability
- Delivery timelines
b) Agent-Friendly Checkout Flows
- Allow verified AI agents to add to cart
- Avoid aggressive CAPTCHAs
- Use clear HTML button labels
- Avoid popups that block progress
c) Machine-Navigable UX
- Predictable DOM structure
- Minimal modal blockers
- Checkout logic visible (not hidden in JS)
6. Strengthen Brand & Entity Signals
LLMs rely on entity understanding to decide which retailers to trust. AI systems evaluate brand authority similarly to search engines.
Actions for Retail Brands
a) Build robust brand/entity pages
- Brand identity & story
- Category strengths
- Trust signals
- Delivery/returns info
- Store or service footprint
- FAQs
- Customer service credentials
b) Strengthen external citations (the new backlinks)
- Press coverage
- Awards & recognitions
- Brand mentions
- Wikipedia-style content
- Reviews in reputable publications
- Editorial features
- Sponsored partnerships
7. Prepare for the New KPI: AI Visibility
Visibility in LLMs is becoming as important as SEO ranking. Retailers must track performance indicators focused on being included in AI-generated recommendations.
Start monitoring AI referrals in analytics – look for parameters such as utm_source=chatgpt.com or other AI identifiers. This helps quantify AI-driven traffic, conversions, and product inclusion frequency.
Three Core KPIs for AI Search:
- LLM Retrieval Rate: How often your products appear in AI shopping answers.
- AI Citation Share: Percentage of AI answers referencing your brand, categories, or content.
- Product Inclusion Frequency: How often SKUs appear in AI recommendations across ChatGPT, Gemini, and Perplexity.
These should be monitored as part of an “AI Visibility Performance Checklist.”