How to Set Up an Online Shop So Google and AI Shopping Agents Can Find, Trust and Sell Your Products (UK 2026 Guide)

More shopping questions are now answered inside an AI interface before anyone visits a shop. Google's AI Mode, ChatGPT and Perplexity all show products, and Google has begun offering a Buy button inside AI Mode to select merchants. If you sell online, your product data is becoming your shop window. This is a practical guide to setting a shop up for that, written for UK sellers, with an honest account of what is and isn't available here yet.

Why bother: the numbers, and how far to trust them

Adobe data, reported by BigCommerce, says traffic to US retail sites from AI sources rose 393% year on year in the first quarter of 2026, and that AI-referred visitors converted 42% better. Those are vendor-reported US figures. AI is still a small slice of most shops' traffic (Chartbeat puts AI referrals at 0.01% of publisher traffic), so treat it as an early, fast-growing channel, not a replacement for Google.

The bigger reason is overlap. Almost everything AI shopping systems need, such as clean product data, accurate stock and prices, structured markup and trustworthy policies, also helps ordinary Google Shopping and search results. You are doing the same work once for several audiences.

The three doors into AI shopping

1. Google, through Merchant Center and its Shopping Graph. Free listings feed Google's shopping surfaces, and per BigCommerce's guide, Shopping ads alone do not qualify you for AI Mode. This is the door that matters most for UK sellers today.
2. ChatGPT, through a product feed you submit to OpenAI. The required fields are item ID, title, description, URL, brand, seller name, image URL, availability and price. OpenAI's standard feed upload currently targets the US, and other markets need OpenAI's confirmation.
3. Perplexity, through a free merchant programme that accepts a Google-format feed, according to BigCommerce. Check Perplexity's current terms.

The UK reality check

• Google's UCP-powered checkout, the Buy button in AI Mode and Gemini, is limited to the US, Canada and Australia and to select merchants for now.
• OpenAI's standard feed upload is US-focused, so a UK seller should not expect to appear in ChatGPT shopping yet.
• That doesn't make this early work wasted. Build the data once, get Google right, and when UK access opens you will already be ready.

The checklist

Work down this list in order. The sections build on each other. Tap any line for why it matters, and your progress is saved in your browser.

Product markup generator

Structured data is how a machine reads your price, stock, delivery and returns without guessing. This generator produces Product, Offer, shipping and returns markup for a UK shop. Paste the result into the product page template, then run the page through Google's Rich Results Test.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Linen Wrap Dress, Sage",
  "description": "Lightweight linen wrap dress with a tie waist and mid-calf hem. Machine washable.",
  "image": [
    "https://example.co.uk/images/linen-wrap-dress-sage.jpg"
  ],
  "sku": "LWD-SAGE-12",
  "brand": {
    "@type": "Brand",
    "name": "Willow & Reed"
  },
  "offers": {
    "@type": "Offer",
    "url": "https://example.co.uk/products/linen-wrap-dress-sage",
    "priceCurrency": "GBP",
    "price": "64.00",
    "itemCondition": "https://schema.org/NewCondition",
    "availability": "https://schema.org/InStock",
    "shippingDetails": {
      "@type": "OfferShippingDetails",
      "shippingRate": {
        "@type": "MonetaryAmount",
        "value": "3.95",
        "currency": "GBP"
      },
      "shippingDestination": {
        "@type": "DefinedRegion",
        "addressCountry": "GB"
      },
      "deliveryTime": {
        "@type": "ShippingDeliveryTime",
        "handlingTime": {
          "@type": "QuantitativeValue",
          "minValue": 0,
          "maxValue": 1,
          "unitCode": "DAY"
        },
        "transitTime": {
          "@type": "QuantitativeValue",
          "minValue": 1,
          "maxValue": 3,
          "unitCode": "DAY"
        }
      }
    },
    "hasMerchantReturnPolicy": {
      "@type": "MerchantReturnPolicy",
      "applicableCountry": "GB",
      "returnPolicyCategory": "https://schema.org/MerchantReturnFiniteReturnWindow",
      "merchantReturnDays": 30,
      "returnMethod": "https://schema.org/ReturnByMail",
      "returnFees": "https://schema.org/FreeReturn"
    }
  }
}
</script>

The golden rule: page, markup and feed must agree

Platforms compare your page, your structured data and your feed. If the page says £64, the markup says £59 and the feed says £64 with "in stock" when you have none, you are teaching every system that your data is unreliable. Three habits prevent most problems:

• One source of truth. Generate the page, the markup and the feed from the same product database. Don't hand-edit three copies.
• Update on change, not on a schedule. When price or stock changes, all three should change within hours. OpenAI's feed reference says to update prices when they change and to update availability the moment an item sells out. Sale dates do not schedule changes for you.
• Out of stock means out of stock. Set the availability field and the markup to out of stock when inventory hits zero.

What the feed needs

OpenAI's published feed specification gives a good picture of what AI shopping systems expect. Google's Merchant Center attributes are similar in spirit.

Use UTF-8, keep IDs as text so leading zeros survive, and never leave the feed to rot: a snapshot at least daily is the sensible floor.

How to structure the store itself

Feeds get you listed. The site itself decides whether you're trusted and quoted. This section is our recommendation, based on how assistants and search systems appear to read pages. It is practice, not an official rule.

Category pages are your buying guides. Open each with two or three sentences that say what the range is, who it suits and what a typical price is. Then show products, then a short FAQ. Category pages are where "best linen dresses UK" searches land.

Product pages answer in this order.

1. A single plain sentence: what it is, who it's for, the price.
2. Price, stock, delivery date and returns, in text.
3. A specification table, with sizes, materials, care and dimensions, as text rather than images.
4. Real reviews, visible on the page.
5. Four or five genuine customer questions with short answers.

Keep URLs boring and permanent. One URL per product, with variants handled through parameters your platform controls and a canonical tag pointing to the main version. Don't change URLs when you rename a product.

Policies are content. Delivery, returns and contact pages should be short, specific, and kept current. Assistants quote policies, and customers check them before they buy.

Publish buying guides that earn citations. "Best X for Y" and "X vs Y" guides that put the answer in the first 100 words, followed by a comparison table, are the pages assistants lift from. Date them accurately, add an "Updated" date only when the content changes, and cite where numbers come from.

Let the right crawlers in

Some shops block these bots by accident. Three things to check:

• robots.txt should allow Googlebot, OpenAI's OAI-SearchBot (used for ChatGPT search and shopping) and PerplexityBot. GPTBot is OpenAI's training crawler, and BigCommerce notes that blocking it doesn't affect ChatGPT shopping. Blocking Google-Extended doesn't remove you from AI Mode or AI Overviews.
• Your CDN or firewall. Cloudflare began blocking AI training and agent crawlers by default on new domains from 15 September 2026, while allowing search crawlers. Look at your server logs to confirm those bots get a 200 response on product pages.
• Rendering. Put price and stock in the HTML the server sends, not only in scripts that run later. A bot that doesn't run your JavaScript sees an empty page.

A 30-minute monthly routine

1. Open Merchant Center Diagnostics and fix any new warnings or disapprovals.
2. Spot-check five products: page price, markup price and feed price all match.
3. Run your top ten buying prompts, such as "best waterproof walking jacket under £150 UK", in ChatGPT, Gemini and Perplexity. Log whether you appear, which product and URL is cited, and whether price and stock are correct.
4. Update one buying guide with something new: a price change, a new product, a fresh review quote. Then update its date.
5. Check your analytics for an AI channel: referrals from ChatGPT, Perplexity and Gemini. The volumes will be small. The trend is what you're watching.

What we'd skip for now

• Agentic checkout. Until it reaches the UK, treat it as something to understand, not to build. The preparation is the same as everything above.
• Gaming tricks. Hidden text, mass-generated product copy and fake reviews are exactly what spam systems target, and Google's September 2026 spam update ran for nearly two weeks.
• Chasing every platform. Get Google's Merchant Center right first. It earns you value today and sets up the rest.

Want it done for you?

We build and manage shops for UK sellers with the product data layer built in: structured data that matches the page, a clean feed, and fresh content. Tell us what you sell and we'll tell you plainly what to fix first.

For the bigger picture on why this matters, read our analysis of what 40% publisher traffic losses and our own Search Console data say about 2026.

Sources

• Get your products into ChatGPT, Google AI Mode and Perplexity, BigCommerce
• OpenAI: product feed reference
• Google Merchant Center Help: Universal Commerce Protocol and UCP-powered checkout
• Google: AI Mode and AI Overviews updates, 15 October 2026
• Reddit and USA Today face Google exit (Cloudflare crawler defaults), PPC Land
• Chartbeat 2026 Publisher Playbook, via Relevant Audience
• September 2026 spam update complete, Relevant Audience