AI will soon be buying on behalf of your customers. What will you do next?

Most discussions about AI and e-commerce focus on discovery: how do customers find your products through ChatGPT or Google Gemini? But something else is happening. AI isn’t just going to make recommendations – it’s going to buy. On behalf of your customer, without them ever visiting your online shop.

That doesn’t just change how customers reach you. It changes what a “new customer” actually means. Because someone who buys through AI doesn’t know your brand, has never seen your website, and barely knows who they’ve ordered from.

The question, then, isn’t whether you’re discoverable by AI. The question is what you do when AI delivers that customer to you.

What is agentic commerce?

The term “agentic commerce” sounds complicated, but the idea is simple: AI doesn’t just make recommendations – it also handles the purchase.

This is now possible thanks to new technical standards: ACP (Agentic Commerce Protocol) and UCP (Universal Commerce Protocol). These are agreements between systems about how they communicate with each other – in this case, between AI platforms and online shops, so that an AI can place an order, request a price and complete a payment.

The result: a customer who buys through ChatGPT has never seen your product page, never looked at a basket, never gone through checkout. AI has done all of that for them.

Will my online shop disappear?

No, but its role will change. Customers who already know your brand will continue coming directly to you. For brand experience, customer service, repeat purchases – your website remains indispensable.

But a growing proportion of your new customers will soon arrive via a route where they’ve never seen your website. They’ve bought through an AI assistant. They barely know your name, let alone your story.

And that’s precisely the point. That first purchase through AI is not a customer relationship. It’s a transaction. Whether you make more of it depends entirely on what you do next.

Four things that are changing structurally

1. Invisible product data doesn’t exist for AI

AI doesn’t look at how beautiful your website is. It reads data: product name, size, colour, price, stock, reviews. Nothing more. Incomplete product data, no size filter, unclear category, outdated stock – that means AI skips your product. Not because it’s poor, but because AI doesn’t understand it. You can have the perfect product and still be invisible.

Example: A customer is looking for a blue raincoat in size M for under £100. You have it. But your product feed only says “raincoat” with no size, specification or price. AI doesn’t show it. Your competitor with complete data wins the sale.

2. An AI customer doesn’t know your brand

With a normal purchase, a customer builds a connection with you along the way: they see your advert, read your reviews, browse your website. By the time they buy, they already know you a little.

With an AI purchase, all of that is skipped. You have an email address and an order. No connection, no recognition, no reason for them to choose you specifically next time.

But this is also your opportunity. The order confirmation is the first real contact with that customer. If that email contains a trackable link, Activate can recognise the customer and enrich their profile, after which Mail+ can start a journey. From a cold transaction to a genuine customer relationship – with a single click.

Example: Customer buys running shoes through ChatGPT. Mail+ sends automatically:

• Day 1: Welcome email with your brand story
• Day 7: Tips for breaking them in + matching socks as a cross-sell
• Day 30: How are you getting on with the shoes?
• Day 90: Reminder that shoes need replacing after approximately 500 km

3. Last-click attribution no longer works

When a customer buys through ChatGPT, you can no longer look back at which advert or keyword made the difference.

What you can measure:

• How do these customers behave after the purchase?
• Do they come back?
• How quickly?
• How much are they worth over a year?

In Activate, you can see this per customer group, so you can compare AI customers with customers from other channels and adjust your approach.

4. Personalisation only begins after the purchase

On your own website, you can make recommendations during the visit. But when the purchase happens through AI, that opportunity isn’t there.

Your personalisation space lies in what comes afterwards: which product fits with the first purchase? When does something need replacing? You play that out through automated email journeys in Mail+. The cross-sell you used to do on the product page, you now do via email – at the right moment, based on what someone has bought.

Six things you can do right now

You don’t have to do everything at once. Start here:

1. Get your product data in order. Complete titles, all attributes filled in (size, colour, material, intended use, etc.). AI can only recommend what it understands.
2. Ensure your prices and stock are up-to-date. AI consults your data in real time. An outdated feed means missed recommendations, even if you have the best offer. Aim for direct synchronisation.
3. Build a welcome journey for people who don’t know you. Set up an automated sequence in Mail+ that tells new customers who you are, what makes you different and what they can expect.
4. Automate your follow-up communication based on the first purchase. Which product fits with it? When does something need replacing? In Mail+, you build this as an automated journey, so every customer receives the right message at the right moment – without manual work.
5. Measure what customers do after the purchase. Not just ‘how many visitors’, but: does someone buy a second time? When? What are they worth over a year? These are the figures that will matter.
6. Centralise your customer data. Purchases, email behaviour, website visits – in Activate, these come together in a single overview. That makes segmenting, comparing and targeted follow-up possible, including for customers who have come in through AI.

The real shift is not technical

Much has been written about the technology behind agentic commerce – the protocols, the integrations, which platform is already ready. That’s interesting, but it misses the point.

The real shift is human. A customer who buys through AI has a different relationship with your brand than a customer who has browsed your website themselves. That relationship simply doesn’t exist yet. You have to build it.

Companies that invest in good product data, smart lifecycle journeys and a central customer view turn every AI transaction into a starting point for a genuine customer relationship. Companies that don’t will remain dependent on the next purchase that AI arranges for them.

And the platform you do it on? That matters less than you think.

Shopify has an early connection to ChatGPT purchases, but the shift applies to everyone. With Magento, WooCommerce, Shopware or PrestaShop, you’ll get just as far – as long as your data and your communications are in order.

Want to know what specifically needs to change for you? We’d be happy to look at it with you.

Frequently asked questions