AI shopping optimisation is the process of making your product catalogue easy for AI assistants (like ChatGPT and Google Gemini) to understand, recommend and link to. It combines clean product data, schema markup, high-quality imagery, trustworthy reviews, and consistent pricing/availability so customers can move from “ideas” to “buy” with fewer steps.

In March 2026, John Lewis made a very public bet on where discovery is heading: away from brand websites as the first touchpoint, and towards AI apps and social platforms where people already ask for ideas.

You don’t need to be John Lewis to respond. But you do need to treat AI assistants as a new discovery layer — one that rewards clean product data, clear content, and a commerce stack that can expose the right information safely.

Why John Lewis’ move matters

When a heritage, omnichannel retailer says its range will be surfaced inside ChatGPT and Google Gemini, it signals three things:

1. “Discovery” is moving upstream. More shoppers start with a question (“What’s a good moisturiser for sensitive skin?”) rather than a retailer name.

2. The winning moment is earlier. If an AI assistant recommends you at the inspiration stage, you enter the shortlist before a traditional search happens.

3. Friction is being designed out. The future experience is fewer clicks from “recommendation” to “purchase”, sometimes inside a third-party app.

So the question becomes: what makes a product catalogue “AI-readable” — and what can you do right now?

What AI assistants need to recommend your products

AI assistants don’t “think” like merchandisers, but they do rely on structured signals. In practice, you’re optimising for three outcomes:

1) Understanding

Your products must be unambiguous.

• Clear titles that match how customers ask questions (“women’s waterproof walking jacket”, not internal naming).

• Consistent attributes (size, colour, material, compatibility).

• Unique identifiers (GTIN/EAN/UPC, MPN, SKU where relevant).

2) Trust

AI tools are conservative about what they recommend when information is thin or conflicting.

• Accurate prices, delivery info, and return policies.

• Real reviews and Q&A that answer common objections.

• High-quality images with descriptive alt text.

3) Actionability

If the user wants to buy, the assistant needs a reliable path forward.

• Product pages that load fast and render well on mobile.

• Stable URLs (avoid breaking links with frequent restructuring).

• Clear availability signals and regional constraints (UK delivery, store stock, exclusions).

The practical checklist: your “AI shopping readiness” score

If you want your products to show up in AI-led discovery, your priorities look surprisingly familiar — but the tolerance for messy data is lower.

Product data hygiene

• Feed completeness: every SKU has price, availability, brand, category, identifiers, images, and key attributes.

• Variant logic: sizes/colours are grouped correctly (avoid duplicate pages that confuse systems).

• Canonical consistency: one preferred URL per product.

On-page structure (PDPs and category pages)

• Schema markup: Product, Offer, AggregateRating where applicable.

• Content that answers prompts: short “why it’s right for you” sections, materials/care, compatibility, and genuine use cases.

• Avoid thin pages: if a product has almost no detail, AI assistants struggle to recommend it confidently.

Inventory and fulfilment signals

• Live availability (in stock / low stock / pre-order) where possible.

• Delivery speed and cost by region.

• Clear returns and warranty details.

Governance and compliance (UK/EU reality)

• Customer data and consent practices that stand up to scrutiny.

• Rules for claims (especially in beauty, health-adjacent categories, and sustainability messaging).

• A process for updating product content when suppliers change specs.

A 90-day roadmap for retail teams

This is the part most organisations miss: AI shopping isn’t a single feature. It’s a coordinated effort across ecommerce, data, marketing, and compliance.

Days 1–30: Fix the fundamentals

• Audit your top 200 revenue-driving SKUs: missing attributes, inconsistent titles, weak images, thin PDP copy.

• Add or validate Product schema across PDP templates.

• Align naming across feeds, PDPs, and onsite search so one product isn’t described three different ways.

Quick win: choose one category (e.g., gifting, skincare, headphones) and make it the “gold standard” for data quality.

Days 31–60: Make your catalogue easier to understand

• Improve attribute depth (materials, dimensions, compatibility, certifications where relevant).

• Expand review capture and on-page Q&A prompts.

• Standardise variants and canonical tags to prevent duplicate content.

Quick win: create a short “best for…” section on PDPs (2–3 lines) written in customer language. That content often maps closely to AI prompts.

Days 61–90: Measure and scale

• Define a tracking approach for AI referral traffic (separate from generic “direct” where possible).

• Build a repeatable content workflow: who owns product truth, how changes are approved, how often pages are refreshed.

• Plan your next integrations (social commerce, marketplaces, partner feeds) based on what your customers already use.

Quick win: publish a category-level buying guide that mirrors AI-style questions. Then link to it internally from PDPs and category pages.

What to watch next (without betting the farm)

You’ll hear a lot about “agentic commerce” — AI systems that can do more than recommend, including taking actions. Whether or not in-app checkout becomes mainstream quickly, the direction is clear: AI discovery is already happening, and it’s compressing the journey.

Your safest investment is still the basics:

• cleaner catalogue data,

• more helpful product content,

• stronger structured signals,

• and better measurement.

That work pays off in Google, in marketplaces, in social commerce, and in AI assistants.

How Generation Digital can help

If you’re building for AI-led discovery, you need both marketing clarity and technical rigour. We help teams:

• Diagnose product data and content gaps that prevent AI tools from recommending you.

• Implement structured data and templates that scale across thousands of SKUs.

• Create buyer-guide content designed for both SEO and AI summaries.

• Build governance so product truth stays accurate as your catalogue changes.

If you’re already investing in AI adoption internally, our playbooks and consulting can help you avoid the “pilot trap” and make improvements that stick.

Next Steps

1. Pick one high-margin category and run the 30-day fundamentals audit.

2. Fix schema and PDP content for the top SKUs first.

3. Set up measurement so you can prove whether AI discovery is driving visits and revenue.

4. Then scale the same approach across the catalogue.

FAQ

Q1: Can customers buy products directly inside ChatGPT or Google Gemini?
In some cases, AI tools are experimenting with shorter paths to purchase. The practical takeaway is to make your catalogue and product pages easy to understand and act on, so you’re ready as these experiences expand.

Q2: What’s the most important technical change for AI shopping optimisation?
Start with clean product data and Product schema on your PDPs. If your identifiers, prices, availability and attributes are inconsistent, everything else becomes harder.

Q3: Is this different from normal SEO?
It’s built on the same foundations, but the content needs to be more explicit and structured. AI assistants summarise and compare, so clarity beats cleverness.

Q4: How do we measure whether AI assistants are sending us traffic?
Track referrals from AI domains where possible, monitor changes in branded search and category traffic, and use landing-page-level analytics to spot patterns.

Q5: What should we do first if we have limited resource?
Optimise your top-selling products: titles, attributes, images, reviews and schema. It’s the highest-leverage work and improves every channel.