Imagine a shopper opens an AI assistant and says she needs a vacuum for a small apartment with a dog. The assistant compares a few options, picks one, and completes the purchase in 9the same chat. 

She never visits a brand site, never sees a product page, and never clicks an ad. A few seconds later, an order shows up in someone's dashboard.

The company that sold the vacuum sees the order. It does not see why that vacuum won, which ones lost, or what the shopper asked along the way.

This is the shift that Google and Shopify set in motion in January 2026 when they announced the Universal Commerce Protocol, or UCP. 

Here, you will get to know what UCP is, what it takes away from brands and retailers, what you can still see, and what to do about it. It follows the analysis by Marta Sulkiewicz of Similarweb, with added explanation and context.

What is UCP?

UCP is an open-source standard that lets AI agents handle a whole shopping trip with any store that supports it. Until now, an AI assistant could suggest products, but the buying itself happened on a retailer's website. With UCP, the agent can do the buying too, inside a conversation in places like Google's AI Mode, Gemini, and ChatGPT.

The protocol covers the full workflow. That includes finding products, checking whether they are in stock, checking out, and handling support after the purchase. 

Before, a merchant would have needed a separate custom integration for every platform that wanted to sell its products. With UCP, a merchant connects once and becomes available to every AI surface that uses the standard.

The timing matters. Gartner predicted back in 2024 that a quarter of search volume would move to AI assistants by 2026. If that is where people ask questions, and UCP lets those same conversations end in a purchase, then a growing share of sales will skip the website entirely.

What Google is Launching First

Google is rolling out three early features built on UCP. The first is Direct Offers, which puts deals in front of a shopper inside AI Mode at the moment she shows she wants to buy. The second is Business Agent, a branded AI representative that can answer questions and finish a sale for a specific company. 

The third is AI Mode Checkout, which lets the shopper pay without leaving the conversation, using Google Pay or another supported method.

The pitch is easy to like: less friction, faster purchases, fewer people abandoning their carts. The catch is less obvious, and it is the reason this article exists.

When You See the Sale, But Not The Decision

When an AI agent buys on someone's behalf, the merchant gets what it needs to fulfill the order. What it does not get is the story behind the choice.

That missing story is the information that brands have always used to grow. You no longer know which competing products the agent looked at. 

You do not know why yours was picked, or why it almost was not. You miss the questions and doubts the shopper raised along the way, and any need she mentioned that your product did not meet. You also cannot tell how much price, stock levels, or brand reputation mattered.

Reviews become a blind spot too. If the agent skips your product, was it because of review sentiment, the price, the stock level, or because your product description did not match what the shopper asked for? 

In the old model, clicks and A/B tests helped you answer that. In agentic commerce, the decision happens inside an AI evaluation you cannot observe or test directly.

BCG has described this shift by saying that your most valuable customer may soon not be a person. Visa's chief product officer, Jack Forestell, has said that people will have AI agents browse, choose, buy, and manage things for them. 

Put those two ideas together, and you get the real risk. Brands and retailers stop being businesses that serve customers directly and start being systems that serve agents. You do not just lose a sales channel. You lose the feedback loop that tells you why you are winning or losing.

Brands and Retailers Play Different Games

UCP does not hit everyone the same way. It creates two separate contests with different rules.

Brands Becoming a Source AI Trusts

For a brand, the contest is no longer about ranking on a results page. It is about becoming a source that AI systems trust enough to recommend.

An agent judges products on how well they match the request, what reviews say about them, and how reliably they are sold, often before the shopper sees any list. Reviews matter more than they used to. They are no longer just proof for human readers. 

AI systems read them to learn how a product performs in real life, what problems come up, and whether it lives up to what the brand claims.

If your reviews are mixed, your quality problems are unresolved, or your positioning is unclear, you can be filtered out early, and nobody tells you why.

That makes several things important. One is how AI agents describe your products and the claims you make. 

Another is whether your brand looks credible and authoritative in AI answers, and whether your message stays consistent across different AI platforms. Search visibility inside your retail partners' websites matters too, as do your price and stock level at the moment the agent checks.

Where the Sale Happens

Winning the recommendation is only half the job. The agent must also pick where the purchase takes place, and that choice affects your margin, your access to customer data, and your long-term position.

Agents choose a channel using price, stock, shipping speed, convenience, and past performance. So brands need to look beyond their own store. Keep your direct-to-consumer site competitive on price, shipping, and inventory. 

Avoid stockouts and price gaps at retail partners. Learn which channels the agents prefer in your category. And weigh the margin you earn on your own site against what marketplaces offer in signals and reach.

A brand that wins on relevance but loses on channel still loses the sale. It may also teach the agents to favor competitors with stronger operations across channels.

Retailers and Marketplaces: Execution Decides

Retailers compete on how well they run their business. Agents will lean toward merchants that reliably turn a shopper's request into a completed order. 

The signals that count are competitive prices and promotions, a wide assortment that fits what shoppers want, products that are actually in stock, fast and dependable shipping, a good range of in-demand brands, and a smooth returns experience with satisfied customers.

In this world, operational failures do not cause a slow decline. They remove you from the picture. A retailer that is overpriced or often out of stock will quietly drop out of AI recommendations. 

That also opens a door for smaller, more flexible retailers to compete against large ones on price, stock, and range.

When Your Brand Competes Against Itself

Here is a question that many brand leaders will have to answer. When an AI agent recommends your product, which version does it buy? The one on your own site, the one on Amazon, the one at Walmart, or the one at a specialist store?

This is already happening, and it changes the economics of brand strategy.

How an Agent Picks the Store

Once your brand wins the recommendation, the agent has to decide where to send the purchase. It weighs price and promotions, such as whether your own site beats Amazon on the deal. 

It weighs shipping speed and cost, and whether a fast-shipping membership program outdoes your standard delivery. It checks which channels have the product in stock, and which store offers the best choice of sizes, colors, or related items. 

It also considers trust and convenience, like whether the shopper already has an account and a saved payment method. Return policies count, and so does the shopper's history of buying from that store.

The Tension This Creates

Many brands have spent years building their direct channel because it brings higher margins and customer data. But an agent might keep choosing a marketplace because of faster shipping and built-in checkout.

You then win the first contest, since the AI chose your brand over others. You lose the second, because the sale lands at a marketplace with higher fees and no customer data for you.

What it Looks Like in Practice

A premium brand might find that agents default to Amazon even when its own site offers exclusive colors or bundles. 

A brand that discounts on its own site might still lose to a retailer that ships faster. Brands that focus only on their own store's conversion rate might find that agents never send anyone there.

So the visibility question gets harder. You need to know whether your brand is being recommended, but also where those recommendations send the purchase, and why.

How to Get Visibility in Practice

UCP may limit the data you receive after a sale, but the contest for visibility is decided before the agent makes its pick. That is where competitive research becomes more useful, not less. The Similarweb article shows what this looks like with its own tools, and the examples are worth understanding whichever tool you use.

Start with AI search visibility. Sparlab AI Search Intelligence shows how a brand appears across ChatGPT, Gemini, Perplexity, and other platforms, using real prompts and real AI answers. 

It tracks brand visibility by topic and competitor, shows which sites and pages influence the AI's recommendations, lists the exact prompts where you appear or do not, and reports the sentiment around each mention. 

In their example, JBL's visibility score for audio equipment is rising, with a strong presence in comparison prompts that signal high purchase intent, while Samsung's is slowly falling. That kind of shift shows up in AI answers well before it shows up in sales.

What to Do Right Now

If You are a Brand

First, find out how AI systems see and describe you today. Ask the main assistants about your products and note what they say and what they get wrong. Then map where those recommendations send shoppers across your own site and your retail partners.

Next, strengthen the content and trust signals that build authority, like clear product claims and healthy reviews. 

Improve your visibility in the on-site search of your retail partners. Compare price, shipping speed, and stock between your own store and your partners, and monitor price and availability for the products that matter most in AI answers.

If You are a Retailer or Marketplace

Decide which AI platforms are worth partnering with. Be honest about where you have an edge in price, availability, and fulfillment. 

Review your content so it is easy for AI systems to read, and compare your visibility with competitors. Make sure you stock the brands that AI agents already favor, and close the gaps that silently take you out of recommendations.

Frequently Asked Questions

What is UCP? 

The Universal Commerce Protocol is an open-source standard announced by Google and Shopify in January 2026. It lets AI agents browse, check stock, check out, and handle post-purchase support with any participating merchant through one common set of rules.

What Data do Merchants Lose? 

They still get order and fulfillment data. They lose the decision data: which competitors the agent considered, why it chose or rejected a product, what the shopper asked or worried about, and how price, stock, and reputation affected the choice.

Can I Still See Anything About AI-driven Visibility? 

Yes, partly. Google Search Console includes AI Overviews and AI Mode in its performance reports, so you can see impressions and positions, including for searches with no clicks. What you cannot do yet is separate AI results from classic results.

Does it Matter Whether I Sell on My Own Site or Through Retailers? 

Yes. The agent chooses where the purchase happens based on price, stock, shipping, convenience, and the shopper's history, and that choice can send the sale to a channel with lower margins and no customer data for you.

Who Wins in Agentic Commerce? 

According to the Similarweb analysis, not simply the companies that integrate with UCP first. The winners will be the ones who keep a clear view of their competitive position, their operational performance, and the factors that drive AI decisions.

Related Searches: