Burak Reis

How a brand can get ready for AI shopping

Before AI can recommend a product, it has to understand it. And it learns that from well-kept product data, not from nicely written product descriptions.

Burak Reis3 min read

While I was building my shopping assistant concept on top of three real stores' interfaces, everything led back to the same place: product data. However smart the assistant is, it can only say what the brand's data says. If nothing anywhere says whether a fabric creases, it can't give an honest answer to someone asking for "a shirt that doesn't wrinkle."

This post is partly for people who aren't designers, the ones on the brand side.

A product page is now a data record

Think of a shopping question: "A linen shirt that doesn't wrinkle, fits true to size, costs under 2,000 TL and arrives by Friday." Every part of that sentence is a separate piece of information: a fabric property, the fit, the price, the delivery time. AI can match them only if they sit in the product data as separate fields. If they're lost inside a nicely written promotional paragraph, it can't.

Most AI shopping tools read products from the structured product feeds that brands send them. You could say the product feed is the shop window that AI sees.

Buying through AI hasn't arrived yet

I should also say there's no need to rush. Discovery has moved to AI; buying, for the most part, hasn't. According to news reports, OpenAI pulled back in March 2026 the direct checkout it had launched inside ChatGPT in September 2025, and handed payment back to brands' own sites.1 Google's in-chat checkout is only available in the US for now.2

For a brand selling in Türkiye or the Gulf, I think the realistic path today goes like this: first, look right in AI's eyes; then, meet the customers who arrive on your own site; and last, build an AI mode on your own site.

You don't need a special file for AI

There's good news too. In a guide published in May 2026, Google says AI features don't need a separate file or any special markup. AI search is still search in the end, and it runs on the same rules.3 So most of what's being sold as "optimization for AI" really comes down to well-kept product data and a decent website.

In practice, the to-do list is plain:

  • Keep each product's attributes in separate fields: fabric, fit, measurements, capacity, care, what's in the set.
  • Keep price and stock identical and up to date everywhere. As I wrote in the previous post, the brand pays for a wrong price.
  • Show product information openly on the page. Don't hide it behind tabs or boxes that only load when you scroll.
  • What others say about you counts too: reviews, news, the listings on marketplaces. Make sure they all agree.

Why I care about this as a designer

When we say brand identity, we think of a logo, colors, typography. But customers now also meet a brand through an AI's answer.

My sense is that over the next few years, good brand work will cover product data alongside the logo.

Notes

  1. Retail Gazette, "OpenAI pulls back from in-chat shopping as ecommerce reality bites", March 2026. Source ↩
  2. Google Merchant Center Help, in-chat checkout. Source ↩
  3. Google Search Central, guide to AI features, May 2026. Source ↩

Published on October 27, 2026.

Burak Reis designs brand identities and motion. About, portfolio.

Follow via RSS