If you have noticed more and more discussion around commerce protocols and agentic checkout, you are not alone. A commerce protocol is a standard that describes how an online store talks to a platform’s services so that orders can be placed at different points of that ecosystem — the Yandex Commerce Protocol (YCP) is one live example of the format.
But something else matters more: a protocol like this is not “one more integration for the sake of it”. It signals a shift that has already begun: choosing and buying increasingly happen where the shopper states their intent — in a conversation with an AI or right inside the search interface. Which means an eCommerce business needs a strategy not only for “how do we get traffic” but for “how do we control the interface where the decision and the purchase happen”.
What a commerce protocol is, in plain terms
A commerce protocol is a standard that describes the exchange between a platform and an online store and makes it possible to place orders across that platform’s services. Once a store supports the protocol, it gets plugged into AI-assistant and search scenarios, and a “Buy in one click” button can appear next to its products.
In practice the key element is the universal checkout: a mechanism that lets the shopper move straight to placing an order from search results or from an answer in an AI chat, without going to the store’s website. The seller is still the online store, and the store receives the order details.
It is worth separating three terms that get mixed up:
- The protocol — the integration standard (what data is exchanged and how an order is placed).
- “Buy in one click” — the shopper-facing button and the fast checkout mechanic behind it; it can appear next to product offers and take the shopper straight to placing the order. In the implementations launched so far it is positioned as free for both shoppers and merchants.
- The merchant console — the service where an online store manages how its products appear in search, submits data via feeds and structured markup, and reads the analytics.
Why this became an eCommerce topic in 2026
Commerce protocols have a combination that is rare on the eCommerce market: they are simultaneously a news event (platforms are announcing them as the standard for agentic commerce) and a practical sales channel that embeds a store into new buying scenarios through search and AI assistants.
The platforms describe them as infrastructure that lets a merchant receive orders which the shopper starts placing directly from an AI chat or from search results, paying in full or in instalments (BNPL). And they stress that the sale itself is made by the online store.
For the market this means that a new point of entry appears, and stores start competing not only over a list of links but over a ready-made shortlist plus a short transactional funnel. This is no longer the classic “search → site → catalogue → filters → product page → cart”.
How general AI is taking traffic and transactions away from online stores
1) AI becomes an “answer engine”: a share of the queries that used to end with a click on a link now end with an answer or a summary inside the platform. Gartner forecast outright that traditional search volume would fall 25% by 2026, with the share moving to AI chatbots and virtual agents.
2) The need to click goes down: if the platform delivers the answer, the comparison or the recommendation inside itself, clicks through to websites can fall — especially at the top of the funnel. In publishing, Digital Content Next recorded a drop in referral traffic from search as AI summaries and AI modes spread; the logic is “got the answer, did not click”.
3) Agentic commerce appears: AI no longer only “advises” — it starts to “do”, from selection through to checkout. McKinsey describes agentic commerce as a scenario where agents can navigate options, negotiate and execute transactions, and puts the global consumer commerce potentially mediated by agents by 2030 at $3–5 trillion.
4) These scenarios are materialising as protocols and standards: the whole point of a commerce protocol is that a store “plugs into” the new buying interface and can receive orders from chat and from search. The mission is no longer “bring the user back to the site at any cost” but “be where the user decides and buys”.
Important: none of this means SEO is dead. It does mean that the share of value is moving from “win the click” to “get into the selection and transaction layer” — and there a store has to think about its data (catalogue, availability, delivery, price), about the interface, and about who controls it.
Why chat became the familiar interface for choosing products
There is a simple reason why “find me something that…” is displacing “I will filter it myself”: shoppers are tired of operating the interface.
When an assortment runs to tens of thousands of SKUs, the classic “catalogue → filters → comparison → reviews” routine becomes hard work. AI chat and AI search promise something else: state your intent (budget, use case, constraints) and the system offers a shortlist and explains the choice. McKinsey describes exactly that trajectory: manual search and comparison gradually give way to a machine-mediated process.
There is survey data confirming the growing role of AI tools in the everyday job of “getting up to speed on something new”. A Russian-market study by OMD and AdIndex (cities of 1M+, respondents aged 18–55) found that actual interaction with AI runs considerably higher than people’s own perception of it, and that among users of AI services, turning to a chatbot has become a visible habit for gathering information and getting into a topic — with the study separately noting the use of assistants and chatbots to search for, research and choose products and services.
Add the expectation that “buying should be fast” and the picture gets stronger: an analytical review by VCIOM reported that a notable share of internet users expect one-click purchases from content and platforms to spread.
How a commerce protocol puts your store inside AI assistants and search
Reduced to practice, a commerce protocol is a bridge between your catalogue and orders and the platform interfaces where the shopper forms an intent and wants to buy.
What connecting actually gives you, at the level a business needs rather than only developers:
A new checkout scenario
Universal checkout lets the shopper move to placing and paying for an order inside an AI chat, with AI agents and in search. The store sells directly and receives the full order details.
“Buy in one click” as a short funnel
Platform documentation describes the button as taking the shopper straight to checkout, with details such as the delivery address filled in automatically from their platform account.
The same documentation notes that the button is in beta, may not be shown for every user and every product, and that its placements include AI chat.
Four integration paths (the reality of implementation)
Important: this is not “one button” but a set of connection paths that depend on the platform your store runs on. The published integration options come down to four:
- stores already running on the platform’s own commerce solution can be connected automatically;
- stores on a mainstream CMS install a ready-made module;
- stores that sell both on their own site and on the ecosystem’s marketplace can switch the functionality on through the merchant console (in that scenario order fulfilment may sit on the marketplace side);
- for other or custom-built CMSs, connection is available over API — often through a beta or a waiting list.
Requirements and limits (so as not to promise the impossible)
Protocols of this kind are currently aimed at stores selling physical goods; digital goods, services and a number of B2B or wholesale scenarios may not qualify.
The underlying merchant infrastructure also expects the store to meet baseline requirements (a working cart, transparent pricing, a site that functions correctly) and to pass quality checks.
Why an external AI channel alone is no longer enough
At first glance a commerce protocol looks like the perfect answer: connect, and your products are already available in AI chat and search, with a convenient fast checkout on top.
Strategically, though, it solves only half the problem: presence at the new point of entry.
The catch is that the external AI channel starts to control:
- how the shortlist is assembled (which products get considered at all, and on what grounds),
- what the comparison interface looks like,
- which attributes count as important by default (delivery, instalments, availability, rating),
- and, increasingly, where the transaction itself happens.
McKinsey warns outright that for many participants in the current shopping flow, the choice between “launch our own agents” and “how do we meet agentic traffic” may become existential. Strategy& makes the same point in its agentic commerce framework, insisting on “agent-ready foundations” and on moving early.
For an online store this translates into plain language: if you do not give the shopper a convenient conversational interface of your own, then even with the protocol connected you risk handing over the external selection layer for good — along with the preference data, the merchandising levers and control over repeat purchases.
Why an online store needs its own AI shopping assistant on site
This is where the “dual strategy” logic closes.
The external AI channel (search and chat) is useful because it brings reach and a new route to an order. But an onsite AI shopping assistant is what lets you:
- help the shopper choose within your assortment and your rules (availability, margin, promotions, A/B policies);
- explain the differences between models and answer questions on specifications, delivery, warranty and compatibility;
- take the load off filters and categories where the shopper does not know “how to search properly”;
- keep the shopper on your domain at the moment they are deciding — and with them, keep control of the experience, the analytics and the future communication.
The argument is further supported by data showing that AI and agents already shape the buying journey through recommendations and conversational support: Salesforce reported that a significant share of online orders during the peak season were “influenced by AI and agents”, including recommendations and conversational support.
If a commerce protocol helps a store be present in an external AI channel, then Gravity AI Shopping Assistant solves the other half of the problem — it gives the shopper the same familiar conversational shopping experience inside the store’s own site.
What an online store should do right now
Turned into a plan, all of the above looks like this:
1) Check whether the protocol applies to your business
If you sell physical goods and your scenario meets the requirements, it is relevant; if you sell services or digital goods, you will need a different strategy.
2) Assemble the data an AI needs to choose from
You need current feeds and structured markup, correct prices and availability, delivery terms and clear product pages — otherwise you will not win in the AI layer even once connected. Merchant documentation singles out feeds and structured markup as the way to get into the index faster and to appear in more placements.
3) Choose your integration path
Work out where you stand: the platform’s own commerce solution, a mainstream CMS module, marketplace plus your own site, or another CMS over API and beta access. That decision sets the timeline, the budget and how the project is organised.
4) Define success metrics for the new funnel
Not just rankings and clicks, but: impressions in product placements, CTR on “Buy in one click”, conversion into checkout, the share of orders coming from the external AI channel, and the effect on CAC and repeat purchases. Merchant consoles are explicitly positioned as analytics tools for presence, traffic and conversion.
5) Design your onsite assistant in parallel
Think about where it will pay off fastest: search, category, product page, cart. The point is not to bolt a chat onto every page but to put it where shoppers have the most questions and where conversion leaks.
If you want to work out how to prepare your store for the new AI-commerce scenario and where an AI shopping assistant would work best — in search, on the product page or in the cart — start with a short review of your site.
FAQ
What is a commerce protocol?
It is a standard describing how an online store interacts with a platform’s services so that orders can be placed across that ecosystem — including inside AI assistant scenarios and in search.
How does it work in AI chat and search?
The key mechanism is universal checkout: the shopper can move to placing an order from search results or from an answer in an AI chat without going to the store’s website; the store receives the order data and remains the seller.
Why should an online store connect to one?
To be present at the new point of entry (AI chat and AI search), to receive orders from those scenarios, and to shorten the path to checkout through universal checkout and “Buy in one click”.
Can a commerce protocol replace an AI assistant on your site?
No, because the protocol solves the external channel problem (presence and checkout inside the ecosystem), while an onsite AI assistant solves the internal one — the selection interface and keeping the decision layer on your own domain. These are different levels of strategy.
How is a commerce protocol different from an AI shopping assistant?
The protocol is an integration standard and a universal checkout inside someone else’s ecosystem. An AI shopping assistant is a conversational interface for selection, comparison and questions that you control on your own site (data, logic, merchandising, analytics).
How does general AI affect eCommerce SEO and traffic?
The focus shifts from “win the click” to “get into the layer of answers, recommendations and comparisons”. Gartner forecast that traditional search volume would be redistributed towards chatbots and agents. In parallel, platforms increasingly answer inside themselves, which reduces the need to click and changes the shape of the funnel.
