Digital Marketing Insights

The buying process online is no longer happening the way it used to. Ecommerce has followed a predictable flow from search to product page to checkout, and marketers have spent years optimizing ads, product pages, and checkout to improve conversion at each step. But lately, that flow has started to change.

Now, a customer can search, choose, and buy a product inside platforms like ChatGPT and Gemini without ever visiting a website. That means the moment where decisions happen is no longer on your site. Interestingly, Adobe reported a 4,700 percent increase in AI driven retail traffic, and Google expects this could reach 15 to 25 percent of ecommerce transactions within the next two years.
This only means one thing if your brand is not set up for this shift, you can expect to miss out on visibility before a customer even reaches your store.

Where This Is Coming From

The reason this is happening is because of something called Google’s Universal Commerce Protocol, or UCP. It is an open standard that lets AI platforms connect with retailers and payment providers to handle the full purchase process.

It was introduced in updates shared on March 19, 2026 by Google’s VP of Merchant Shopping, and most people have not paid attention to it yet. UCP is built with companies like Shopify, Walmart, Etsy, Wayfair, and Target, with support from Visa, Mastercard, and Stripe. Instead of sending users to a website, it allows the AI to pull from a retailer’s product data, including real time pricing and inventory, build a cart, and complete the transaction in the same interaction.

This is already happening in real time for retailers, and was available earlier in Gemini before expanding further. For example, a shopper can open Gemini, ask for a kitchen setup under a set budget, and the system can pull available products from a retailer, add them into a cart, and complete the purchase without the user ever visiting the store’s website.

The Updates That Made This Work

At launch, this system was limited. It could only handle one item per transaction, which does not reflect how people actually shop. It also relied on Merchant Center feeds that update on a schedule, which meant pricing and inventory could be out of date. A product could show as available even if it sold out earlier that day.

The March 19, 2026 update fixed these issues:

Cart

It can now build a full cart from a single retailer instead of just one item. For example, instead of buying just sunscreen, a customer can add a cleanser, moisturizer, and serum from the same store in one purchase.

Catalog

What it does is it pulls real time product data, so pricing, stock, and variants are accurate. If a product on Shopify or Walmart is sold out or discounted, the system reflects that immediately instead of showing outdated information.

Verification

It checks availability and price before completing the purchase. Before checkout, the system confirms the item is still in stock and the price has not changed since it was added to the cart.

If someone asks for tools for a DIY project under a set budget, the system can pull multiple products from a retailer, like a drill, screws, and safety gear, add them into one cart, and complete the purchase in one flow. This moves it from a limited demo into something that matches how ecommerce actually works.

What This Means for Brands

This changes what determines whether your product gets seen at all. Platforms like Gemini are pulling from your product feed in Google Merchant Center and checking it against your product pages to decide what to show. There are three things brands need to get right:

Your product data needs to be complete and specific

Platforms are reading your titles, descriptions, and attributes to decide if your product matches a request. If key details like size, material, or use case are missing, the system cannot confirm the product fits and it gets skipped. This applies across retailers using platforms like Shopify, Walmart, Etsy, and Wayfair.

Your data needs to be accurate in real time

UCP now pulls live pricing, stock, and variants. If your product is out of stock or discounted on your store but your feed is outdated, it creates friction and reduces visibility. The system checks this before completing the purchase, so errors here directly affect whether your product is shown.

Your product page needs to match your feed

Google compares your Merchant Center feed with your product page. If your feed says one thing and your page says another, it gets flagged. Schema markup also needs to clearly define price, availability, and reviews so the system can understand the product without guessing.

This is already happening in real time. The infrastructure is live, the partners are connected, and transactions are already taking place through systems supported by companies like Google, Shopify, and Stripe. Brands that treat their product data seriously now have an advantage while this is still early and not yet crowded.