Agentic commerce is the most consequential shift in online retail since mobile, and most UK brands are not ready. AI agents now research, compare, and increasingly purchase products on shoppers’ behalf, often without ever loading your website. ChatGPT’s Instant Checkout went live in September 2025. Google’s Universal Commerce Protocol followed in January 2026 with Walmart, Target, Shopify and 20+ enterprise partners. McKinsey projects $900 billion to $1 trillion in US agentic commerce revenue by 2030. This is the no-fluff, UK-focused playbook for what agentic commerce actually is, the protocols that matter, and what your store needs to do, this quarter, to avoid becoming invisible to AI agents.
What Is Agentic Commerce? A Plain-English Definition
Agentic commerce is online shopping where an AI agent does the buying, not the human. The shopper tells an AI assistant what they want, “find me the cheapest non-stop flight to Berlin landing before noon”, “reorder my coffee beans every three weeks”, “build me a taco-night cart from my local supermarket”, and the agent executes the entire purchase: finding products, comparing options, applying discounts, and completing checkout, often without the shopper ever visiting a merchant’s website.
It’s the difference between AI helping you shop and AI shopping for you. Stripe defines it as a system where the customer states their goal in natural language and an autonomous agent handles every step of the transaction in their place.
How it differs from regular AI shopping assistants
Most ecommerce brands have used some form of AI for years: chatbots, recommendation engines, AI search. Those are reactive, they respond to a prompt, surface options, and then a human still drives the purchase. Agentic commerce flips that script. The agent is proactive. It plans, decides, and acts independently within the bounds the user has set. Salesforce frames the distinction simply: bots are reactive, agents are proactive.
The new shopper journey
- Intent expressed in natural language, not entered as a search query.
- Agent searches agent-friendly product catalogues, not your website.
- Agent evaluates based on structured data: price, attributes, reviews, availability.
- Agent transacts via a checkout protocol, often inside ChatGPT, Gemini or another AI surface.
- Merchant fulfils the order, you remain the merchant of record but the shopper never visited your site.
Some publications and vendors use the shorthand “a-commerce” or “agent commerce” interchangeably with agentic commerce. They mean the same thing. The term “agentic” emphasises the autonomous, goal-driven nature of the AI agent doing the buying.
Why Agentic Commerce Matters Now (Not in 5 Years)
For the past two years, “AI in retail” has been mostly hype, content, generation, recommendation engines, marginal personalisation gains. Agentic commerce is the moment that hype turns into infrastructure with real money flowing through it. Three things changed in the last twelve months.
ChatGPT Instant Checkout went live (September 2025)
OpenAI launched its Agentic Commerce Protocol (ACP) in production with Etsy and Shopify merchants. Buyers can now complete purchases inside ChatGPT without leaving the conversation. OpenAI charges merchants a 4% transaction fee. ChatGPT has approximately 900 million weekly users.
Google’s Universal Commerce Protocol launched (January 2026)
At NRF 2026, Google announced UCP with Shopify, Walmart, Target, Home Depot, Lowe’s, Best Buy, Macy’s, Visa and Mastercard backing it. The “buy” button is coming directly into AI Mode in Google Search and into Gemini, the same surfaces hundreds of millions of UK consumers already use daily.
Adoption hit a tipping point
Capital One Shopping data shows 61% of consumers have used GenAI tools to shop and 80% plan to in 2026. Adobe Digital Insights tracked a 4,700% YoY surge in AI-referred traffic to retail sites by July 2025. AI-referred shoppers convert 31% higher than traditional channels and spend 45% more time on site.
Unlike VR shopping or NFT marketplaces, agentic commerce has working infrastructure, transacting users at scale, and meaningful merchant fee revenue today. McKinsey, Morgan Stanley and Forrester are all forecasting hundreds of billions in volume by 2030. The behaviour change is already happening. The only open question is which merchants get found by agents and which stay invisible.
How Agentic Commerce Actually Works
Most explainer articles stop at “an AI buys things on your behalf.” That’s true but useless if you’re the merchant trying to figure out what to do. Here’s the technical reality, end to end.
The agentic commerce stack in 2026
- The intent layer. Shopper expresses a goal to an AI surface, ChatGPT, Gemini, Perplexity, Claude, or a brand-specific assistant. Natural language, no keywords.
- The reasoning layer. The LLM (GPT-5, Gemini, Claude, or similar) parses intent, plans the multi-step task, and decides what tools and data sources to use.
- The discovery layer. The agent queries merchant catalogues exposed via a protocol, ACP, UCP, or MCP, looking for products that match. Your structured data, not your homepage, is what gets read.
- The decision layer. The agent evaluates options against the user’s stated criteria, brand preferences, budget, delivery date, applies any saved preferences, and selects.
- The transaction layer. Payment flows via a token-based system (e.g. Stripe’s Shared Payment Tokens) where the agent never sees the card. The user pre-authorises spend; the agent initiates the transaction; you remain merchant of record.
- The fulfilment layer. The order lands in your normal commerce backend exactly like any other order. The shopper gets confirmation in their AI chat, not via your typical order-confirmation email flow.
If your product catalogue isn’t exposed via a protocol, agents can’t find your products at all. If your structured data is incomplete, agents skip you for competitors with cleaner data. If your inventory isn’t real-time, agents build carts that fail at checkout, and the experience degrades for the shopper, who blames you, not the agent.
The Three Protocols Powering Agentic Commerce: ACP vs UCP vs MCP
Three protocols dominate agentic commerce in 2026, each backed by a different ecosystem. They’re not mutually exclusive, most serious merchants will need to support multiple, the same way you ran Google Ads and Facebook Ads in parallel for the last decade.
The protocol behind ChatGPT Instant Checkout. Merchants expose product catalogues that ChatGPT can discover and transact with. Stripe handles the Shared Payment Token layer. Live since September 2025; ~1M Shopify merchants integrating.
Google’s answer to ACP. Powers shopping inside Google Search AI Mode and Gemini. Co-developed with Shopify; endorsed by Walmart, Target, Visa, Mastercard, Home Depot. Launched January 2026 at NRF.
The general-purpose tool-connection standard, originally built by Anthropic, now governed by the Linux Foundation. Used by Claude, Salesforce and Adobe for catalog connections. The plumbing layer underneath many agentic systems.
Quick comparison table
| Feature | ACP | UCP | MCP |
|---|---|---|---|
| Primary surface | ChatGPT | Google AI Mode + Gemini | Claude + many agents |
| Live since | Sept 2025 | Jan 2026 | Late 2024 |
| Main backer | OpenAI + Stripe | Google + Shopify | Anthropic + Linux Foundation |
| Merchant fee | 4% per transaction | Not yet disclosed | None (open standard) |
| Best for | Reaching ChatGPT’s 900M+ weekly users | Capturing high-intent Google search users | Connecting catalogues to multiple agent systems |
| Fits with Shopify | Yes (rolling out) | Yes (co-developed) | Yes (via integrations) |
Treat ACP and UCP as parallel, not competing. ACP gives you reach into ChatGPT’s research/discovery audience. UCP captures high-intent Google searchers. MCP is the underlying plumbing your tech stack will increasingly speak. Smart UK merchants are scoping all three, prioritising ACP first (it’s already live and Shopify integration is rolling out fastest), with UCP next.
2026 Agentic Commerce Statistics: What the Data Actually Shows
There’s a flood of vendor-driven hype around agentic commerce. Here’s the picture from credible third-party research, McKinsey, Morgan Stanley, Adobe, Capital One Shopping, and the peer-reviewed Kaiser & Schulze study covering 973 retailers and ~$20 billion in revenue.
Adoption and traffic
| Metric | Figure | Source |
|---|---|---|
| Consumers who’ve used GenAI to shop | 61% | Capital One Shopping (2026) |
| Consumers planning to use GenAI for shopping in 2026 | 80% | Capital One Shopping (2026) |
| YoY growth in AI-referred retail traffic (July 2025) | 4,700% | Adobe Digital Insights |
| Higher CVR from AI-referred shoppers vs. traditional | +31% | Adobe Analytics 2025 holiday data |
| Higher revenue per visit YoY from AI-referred | +254% | Adobe Analytics 2025 holiday data |
| More time on site from AI-referred shoppers | +45% | Adobe Analytics 2025 holiday data |
| Lower bounce rate from AI-referred shoppers | -33% | Adobe Analytics 2025 holiday data |
Market projections
| Forecaster | Forecast | Timeframe |
|---|---|---|
| McKinsey | $900B-$1T US, $3-5T globally | By 2030 |
| Morgan Stanley | $190B-$385B (10-20% of US ecommerce) | By 2030 |
| Bain | $300B-$500B (15-25% of US online retail) | By 2030 |
| Gartner (B2B) | 90% of B2B purchases handled by AI agents | By 2028 |
| Forrester | 20% of B2B sellers facing agent-led negotiations | By end of 2026 |
“AI-referred traffic is growing fast and converts better than paid social. It is also still below organic search and affiliate on every conversion metric. The pattern looks like research-stage behaviour, long sessions, deep browsing, lower conversion, not purchase-stage. Parity with organic search inside a year is unlikely.”
Paraphrased from Kaiser & Schulze, peer-reviewed study (SSRN 5585812, October 2025)
The honest read: agentic commerce is moving fast but isn’t yet replacing search or affiliate. It’s a parallel channel, growing exponentially, with conversion rates that will close the gap as merchant infrastructure matures. The question isn’t whether to invest, it’s how soon.
Agentic Commerce Readiness Checker
Answer six questions to see how prepared your store is for AI agents in 2026. The score reflects how findable, parseable and transactable your catalogue is to an autonomous agent today.
How agent-ready is your store?
Choose the closest answer for each question. Takes 60 seconds.
out of 100
How to Prepare Your Store for Agentic Commerce
The merchants who win in agentic commerce aren’t the ones with the most innovative tactics, they’re the ones with the cleanest foundational data. Most of this work isn’t novel; it’s the infrastructure your store should already have, applied with new urgency. Here’s the priority order.
Audit and complete your structured product data
Every product needs accurate GTINs, SKUs, complete attributes (size, colour, material, dimensions), high-quality images with proper alt text, and full descriptions. Mirakl research shows 42% of customers abandon purchases due to insufficient product information; agents apply that same logic at scale and skip incomplete listings entirely.
Implement comprehensive Schema.org markup
At minimum: Product, Offer, AggregateRating, Review, Brand, and BreadcrumbList. Use Schema.org’s Product specification as your blueprint. Validate with Google’s Rich Results Test. Schema is what allows AI agents to understand your inventory in a way they can act on, not just describe.
Connect to ACP if you’re on Shopify
Shopify’s ACP integration is rolling out across merchants. If you’re a Shopify or Shopify Plus store, register interest, prepare your product data, and ensure your checkout supports the Stripe Shared Payment Token flow. Brands like Glossier, SKIMS, Spanx and Vuori went live as early partners. Our UK Shopify team handles ACP setup as part of agentic-readiness work.
Plan for UCP coming through Google
UCP launched at NRF in January 2026 with Shopify, Walmart, Target and others. If your store is already in Google Merchant Center with strong product feeds, you’re already further along than most. Keep feeds healthy and watch for UCP onboarding announcements through Google Cloud and your platform.
Move from static feeds to real-time inventory APIs
Agents that build a cart and find one item out of stock will abandon the entire transaction. Real-time inventory exposed via API isn’t a nice-to-have anymore, it’s a baseline requirement for serious agentic commerce participation. If your inventory updates nightly via spreadsheet upload, that needs fixing.
Build attribution for AI-referred traffic
Most merchants can’t yet distinguish ChatGPT-referred traffic from organic search in their analytics. Set up UTM tracking for AI sources, use referrer logs, and consider tools like the new generation of AI-channel analytics platforms. You can’t optimise what you can’t measure.
Want a 90-day agentic readiness roadmap?
5MS runs an agentic commerce audit specifically for UK ecommerce brands. We map your current state to ACP, UCP and MCP requirements, prioritise the work, and execute the rebuild. No fluff, no consultancy theatre.
Schema Markup & Product Data: The Agentic Commerce Foundation
If you only do one thing this quarter to prepare for agentic commerce, make it this. Structured product data is what AI agents read; everything else is theatre. Here’s the minimum viable schema implementation for any UK ecommerce store.
The agentic commerce schema checklist
- Product schema implemented on every product page (name, description, brand, image, SKU, GTIN/EAN/MPN)
- Offer schema on every variant (price, priceCurrency in GBP, availability, itemCondition, priceValidUntil)
- AggregateRating schema with realistic ratingValue and reviewCount
- Review schema for individual reviews (author, datePublished, reviewBody, reviewRating)
- Brand schema linked correctly to your registered brand entity
- BreadcrumbList schema for category navigation context
- FAQPage schema on product pages with common purchase questions
- Organization schema sitewide with logo, sameAs links, contactPoint, address
- Validated through Google’s Rich Results Test with no errors or warnings
- GTINs/EANs accurate and matching manufacturer data (don’t invent these)
- Real-time inventory feeding the availability field, not stale data
- High-quality product images at minimum 1200x1200px with descriptive alt text
Stores frequently ship Schema.org markup with placeholder values, “5 stars” hardcoded into AggregateRating, dummy GTINs, identical descriptions across variants. Agents read this and either skip your products or, worse, surface them with bad data that frustrates the shopper. Invalid schema is worse than no schema. Validate every template before deploying.
Beyond product schema: the broader data picture
Schema is necessary but not sufficient. The full agentic-readiness data picture also includes:
- Product feed for Google Merchant Center, kept healthy, with all attributes populated, the foundation UCP will build on.
- Stable, descriptive URLs per product. Agents prefer human-readable, persistent URLs over query-string hashes.
- OpenAPI documentation for any public catalogue endpoints, increasingly relevant as agents start querying merchants directly via MCP.
- Up-to-date sitemap.xml and robots.txt, including signals for AI crawlers (e.g., explicit allow rules for
GPTBot,ClaudeBot,PerplexityBotif you want them indexing you).
Answer Engine Optimisation: SEO’s Successor
If agentic commerce is the new buying surface, Answer Engine Optimisation (AEO) is the new SEO. Traditional SEO optimises for ranking in Google’s blue links. AEO optimises for being cited, summarised and recommended inside AI-generated answers, ChatGPT responses, Gemini summaries, Perplexity citations, AI Overviews in Google.
The shift is significant: by 2026 Google’s AI Overviews already cover an increasing share of commercial queries, and analyst estimates suggest AEO will replace traditional keyword SEO for roughly 30% of retail queries within the next 18 months.
What AEO actually means in practice
Write content that answers questions cleanly
AI summarisers extract clear, complete answers, not keyword-stuffed prose. Lead each section with a direct answer in 40-60 words, then expand. Use the question itself as the H2 or H3 where natural. This is exactly what Google’s FAQPage schema rewards.
Cite authoritative sources visibly
AI agents trust pages that cite reputable third parties. Linking to McKinsey, Stripe, Adobe, GOV.UK or major news outlets where relevant gives the AI evidence to summarise from. Pages that read like first-person assertions without sources get cited less.
Build topical depth, not isolated articles
AI summarisers pull from sites with comprehensive coverage of a topic. One article on “agentic commerce” doesn’t get cited; a site with twelve interconnected articles covering every facet of the topic does. This is where content strategy stops being a marketing nice-to-have and becomes core revenue infrastructure.
Create FAQ blocks worth quoting
AI agents love clean, factual question-answer pairs. Build FAQ sections on every product page (sizing, materials, returns, warranty), every category page (buying guidance), and every blog article. This is the highest-yield AEO work most merchants ignore.
Don’t tear up your SEO strategy. Traditional organic search is still the largest single referral source for ecommerce, and that won’t flip overnight. The right move is parallel investment: keep doing keyword SEO well, and add AEO discipline on top. The two reinforce each other, both reward clean, structured, authoritative content.
Agentic Commerce: UK-Specific Considerations
Most agentic commerce coverage is US-centric, and the protocols themselves rolled out into US checkouts first. UK merchants have a slightly different set of considerations, opportunities and obligations.
What UK merchants should know
UK adoption is lagging US adoption by 6-9 months
ACP and UCP rolled out into US merchants first. Most major partner brands announced have been US (Walmart, Target, Home Depot, Lowe’s, Macy’s, etc.). UK retail is integrating more slowly, which is both a problem (you risk being late) and an opportunity (early UK adopters get an outsized first-mover advantage in the UK market).
GDPR and UK data protection apply to agentic transactions
An AI agent acting on a shopper’s behalf doesn’t dissolve your UK GDPR obligations. You still need a lawful basis for processing, you still need to honour data subject rights, and your privacy policy needs to address agentic commerce explicitly. The Information Commissioner’s Office has not yet issued specific agentic guidance, but existing principles apply.
UK consumer protection law still applies, including returns
The Consumer Rights Act 2015 and Consumer Contracts Regulations 2013 don’t have an “AI agent” exemption. If an agent buys on a consumer’s behalf and the consumer wants to return the item under their statutory 14-day right, that right applies. Your returns infrastructure has to handle agent-initiated orders the same way it handles direct ones.
VAT and pricing transparency
UK merchants must continue to display prices inclusive of VAT to UK consumers, including in AI shopping experiences. Your product feed should specify priceCurrency: GBP and ensure VAT-inclusive pricing flows correctly into agentic surfaces. Get this wrong and shoppers see surprise costs at checkout, the fastest way to a poor agentic experience review.
Royal Mail and UK fulfilment integration
Most agentic commerce demos feature US-centric shipping (USPS, FedEx Ground). UK shoppers expect Royal Mail tracked delivery, DPD, Evri, or similar. Make sure your shipping options surface correctly through whatever protocol you support, and that delivery promises are accurate. Agents will optimise for “fastest delivery”; if your data is wrong, they’ll skip you for a competitor.
Because UK adoption is lagging, the merchants getting agentic-ready in 2026 face less competition for AI-agent visibility than US peers. By the time the bulk of UK retail wakes up to agentic commerce in 2027, early adopters will already have established positions in the discovery flow. This is the 2010 mobile-commerce moment for UK ecommerce, the brands that prepared early dominated for the decade after.
10 Agentic Commerce Mistakes UK Merchants Are Making
These are the patterns we see in most UK ecommerce stores when we audit them against an agentic readiness checklist. None are catastrophic individually; together they make a brand functionally invisible to AI shopping agents.
Treating it as “future work”
ACP went live in September 2025. UCP launched January 2026. The brands waiting for “the technology to mature” are missing the actual rollout window.
Incomplete or invented GTINs
Agents cross-reference GTINs against manufacturer data. Invented or duplicated GTINs cause your products to be skipped or flagged. Source real GTINs from your suppliers.
Static, daily-batched inventory
Agents need real-time. If your inventory updates once a day, every cart they build risks failing at checkout for stock that sold an hour ago. Move to real-time APIs.
Schema markup with placeholder data
Hardcoded “5.0” ratings, dummy review counts, default Brand entries. Agents detect this and downgrade trust accordingly. Use real, validated data.
Blocking AI crawlers reflexively
Some merchants have blocked GPTBot, ClaudeBot, and PerplexityBot in robots.txt out of an outdated 2023 instinct to “protect content from AI”. For ecommerce, this is now equivalent to blocking Google. Allow the crawlers you want surfacing your products.
Thin product descriptions
Agents need detail to evaluate fit. “Comfortable, stylish hoodie” tells an agent nothing. Materials, fit, sizing, washing care, weight, fabric weight in GSM, country of origin, sustainability credentials, all of it matters.
No FAQPage schema or AI-shaped Q&A content
The single highest-yield AEO investment, and most stores haven’t done it. Add FAQPage schema to product pages and key articles.
Treating the 4% ACP fee as the whole picture
OpenAI’s 4% ChatGPT fee sounds steep next to a 0% organic referral. But ChatGPT-referred AOV and CVR are higher, with longer time-on-site post-checkout. Calculate true contribution margin per channel before refusing on principle.
No analytics on AI-referred traffic
If you can’t see what AI is sending and what it’s converting, you can’t optimise. Set up source tracking, build dashboards, and review monthly.
Not engaging an agency that understands this
Most generalist ecommerce agencies are still learning. Agentic readiness work needs a team that understands schema, protocols, the UK regulatory layer, and modern Shopify infrastructure. Our team has been operating in this space since the protocol launches and runs agentic-readiness as a defined service.
Agentic Commerce in 2026: The Short Answer
Agentic commerce is online shopping where AI agents discover, compare and buy products on shoppers’ behalf, often without the shopper visiting a merchant’s website. It’s powered by three main protocols: ACP (OpenAI/Stripe, live in ChatGPT), UCP (Google/Shopify, live in AI Mode and Gemini), and MCP (Anthropic, the underlying tool standard). To prepare a UK ecommerce store, complete your Schema.org product markup, expose real-time inventory via API, register for ACP and UCP onboarding through Shopify, build AEO-optimised content, and track AI-referred traffic in analytics, in that priority order.
The 10-point UK action list:
- Audit your structured product data, GTINs, attributes, descriptions, images.
- Implement complete Schema.org markup on every product page.
- Validate with Google’s Rich Results Test, no errors or warnings.
- Move inventory from batch to real-time API.
- Connect to ACP via Shopify if you’re on the platform.
- Prepare for UCP rollout by keeping Google Merchant Center feeds healthy.
- Allow AI crawlers in robots.txt (GPTBot, ClaudeBot, PerplexityBot).
- Add FAQPage schema to every product and key article.
- Build AI-traffic tracking in analytics with proper attribution.
- Run a full agentic-readiness audit, ideally before the end of Q2 2026.
Ready to become agent-ready before your competitors?
5MS is a UK ecommerce agency operating at the front of the agentic commerce shift. We run readiness audits, schema rebuilds, ACP/UCP onboarding, and AEO content strategy as integrated work, not separate workstreams. Book a free 30-minute call to see where your store stands.
Agentic Commerce: Frequently Asked Questions
Sources & further reading
- McKinsey: The agentic commerce opportunity (October 2025)
- Stripe: 10 things we learned building agentic commerce (March 2026)
- Stripe: Agentic Commerce Guide (2026)
- Google Cloud: A new era of agentic commerce is here (January 2026)
- Salesforce: What Is Agentic Commerce? (2026)
- Forrester: Is Agentic Commerce A Thing? (October 2025)
- OpenAI Developers: Agentic Commerce Protocol documentation
- Capital One Shopping: AI Shopping Statistics 2026 Report
- Adobe Digital Insights: 2025 Holiday Shopping Analysis
- Kaiser & Schulze: peer-reviewed agentic commerce study (SSRN 5585812, October 2025)
- Schema.org: Product, Offer, AggregateRating specifications
- Information Commissioner’s Office (ICO): UK GDPR guidance
Data verified April 2026. The agentic commerce landscape is moving rapidly; this article is updated quarterly. Bookmark and return for updated benchmarks and protocol developments.