Only 30 Shopify Merchants Used OpenAI’s Instant Checkout. Here’s Why That Matters.

You know, out of millions of Shopify merchants, only about 30 went live with OpenAI’s Instant Checkout. That was the whole launch. Six months ago, the industry told you that ChatGPT was about to kill the retail website, and today I want to walk you through what actually happened, why most of the takes you’re seeing miss the point, and exactly what to do about it wherever you are in your entrepreneurial journey. Whether you’re doing $500K a year or $50 million a year, I want to walk you through all of that.

For context, about four weeks ago, I recorded episode 453 and I told you that Shopify’s new agentic storefronts were a distribution shift you didn’t want to sit out. I still stand by that, but the story has gotten significantly more nuanced in the last four weeks since I recorded that episode, and I wanted to give you a more updated read on what’s going on.

In this episode, I’m going to walk you through what really happened with Instant Checkout, why most people got the framing completely wrong, what’s actually true about agentic commerce as it is right now in April 2026, and what to do about it depending on your stage. This isn’t a doom take, and it’s not a cheerleader take.

This is an operator read. I want to talk to you as if you were sitting across the table from me at a mastermind.

So let’s talk about what actually happened and get some better context.

Around September of 2025, OpenAI launched what they called Instant Checkout inside ChatGPT. The launch partners were Shopify, Etsy, and Walmart. Even Shopify’s president, Harley Finkelstein, called it “the new frontier of online retail,” and the press loved it. Vendors started pitching “agentic commerce readiness” to merchants as an urgent, important initiative that needed attention right away. The framing everywhere was that buying products inside a chat window was about to replace retail websites.

Here we are six months later, and here’s the actual scoreboard: out of millions of merchants on Shopify, only about 30 ever went live with this feature. That number has been confirmed directly with Forrester. I know some reporting put it closer to 12, but either way, adoption was minimal. Walmart had around 200,000 products available on launch day, but consumers weren’t really buying through it.

What they were doing, like most of us do, was researching in ChatGPT and then completing the purchase on the merchant’s own website, where they had saved payment methods, account history, loyalty memberships, and rewards. All of that lives on the merchant’s website.

The early warning signs came around March 4th. Walmart’s Executive Vice President of AI Acceleration, Daniel Danker, was at a Morgan Stanley tech and media conference, and he flat-out called Instant Checkout “a very temporary moment in time.” The very next day, the information broke that OpenAI was going to kill this feature. On March 24th, OpenAI officially confirmed it through a blog post.

That same day, Shopify announced its Agentic Plan, now globally available, with checkout moving back to the merchant’s own site—which I agree with—and products staying discoverable in chat.

Here’s the kicker that really tells you how unready this product was. As of February 2026, OpenAI had not built the infrastructure to collect US sales taxes. They launched a national checkout product without solving for sales tax. That makes no sense, and it’s not a minor oversight. It’s a foundational gap that should have been addressed from day one. That kind of detail explains a lot about why only around 30 merchants ever went live with the product.

The product clearly was not ready, and the people closest to it likely knew that. So that’s what happened. The hype said retail websites were dead. The reality was that the most ambitious agentic commerce launch in history shipped, struggled, and got pulled six months later.

The best way to understand what went wrong is to talk about the framing. The pitch was that AI would replace the retail website, but that was never how this was going to play out. The data now is proving that.

Forrester ran a consumer research survey in March 2026 and found that completing a purchase inside an answer engine like ChatGPT or Perplexity is the least adopted use case among regular users of those platforms. The top two use cases were exactly what you’d expect—and what you and I already do: asking general questions and researching products. It’s all part of the discovery and research phase.

Buying directly in the chat? Most people are saying no to that, and that’s a big reason the feature was pulled.

There’s also a more structural reason behind this. Once you see it, the whole story starts to make a lot of sense.

Full-stack agentic commerce—where you do everything from discovery to purchase to fulfillment in one AI experience—actually works in one place in the world right now: China. It’s Alibaba’s Quen app, and it works at scale. People are ordering food, booking travel, and buying products inside a single conversational interface, a super app.

It works there because Alibaba owns the AI model, the marketplace, the payment rails, and the logistics. They own the whole stack.

OpenAI tried to do the same thing in the United States without owning any of it. It broke on almost every dimension: tax collection, real-time inventory, multi-item carts, loyalty, customer trust—everything.

I want to be honest about what I think is the most important part of this conversation.

I’ve spent nearly 20 years in ecommerce, six and a half of those inside Shopify, and I’ve watched hundreds of merchants make implementation decisions in real time. I want to name a pattern that triggers every alarm bell I have: shiny object syndrome—the shiny object pattern.

I see it a lot in brands that have product-market fit and are doing somewhere between $500K and $2 million. A vendor announces something new, the press calls it revolutionary, and suddenly merchants are told they need to invest engineering hours, marketing budgets, and operational attention into it right now.

The merchants who actually did the smart thing were not the ones who chased the announcement. They were the ones who kept building fundamentals while everyone else was distracted by the shiny object.

Four weeks ago I told you to pay attention to agentic storefronts, and I still stand by that. But I want to refine that a bit. The version of agentic commerce that’s actually going to matter looks very different from the version we were being sold six months ago. It’s worth getting a clear read on that before you spend a single hour or dollar on it.

Here’s the strategic insight I want you to internalize because it does most of the work for understanding everything that comes next:

Agentic commerce is splitting cleanly into two layers, and only one of them is consolidating in AI.

The first layer is discovery. Discovery is where AI is winning right now, and it’s winning fast. Adobe’s data shows that AI traffic to US retail sites grew 4,700% year over year. AI referral traffic was up over 600% during Black Friday and Cyber Monday in 2025. Shopify’s own AI-attributed orders grew 11x in the last 12 months, and AI-referred shoppers convert 31% to 42% higher than human shoppers from traditional channels.

These aren’t hype numbers. They’re real, compounding numbers that are changing how people find products.

The second layer is the transaction layer. Transaction is where merchants win, and it’s not even close. The trust lives on your storefront. Your saved payment methods live on your storefront. Your loyalty memberships, post-purchase support, returns, customer data, and merchant-of-record obligations all live with you, the merchant.

Trying to reallocate that into a chat window is why Instant Checkout broke on every dimension. OpenAI tried.

So what’s actually shipping right now, in April 2026, is quite a lot—and most of it is getting ignored by the headlines.

If you’re a Shopify merchant, agentic storefronts are now turned on by default for eligible stores. Your products get surfaced right now in ChatGPT, Microsoft Copilot, Google AI Mode, and Gemini. The customer completes their purchase on your storefront—in a mobile browser, tablet, or desktop. You own the customer relationship, and the orders flow into your Shopify admin like usual, with referral attribution so you can see where those sales come from.

Shopify went live with its Agentic Plan globally on March 24th. That part didn’t get nearly enough attention. Brands not on Shopify can now upload their products to Shopify Catalog and distribute through the same AI channels, paying Shopify for the distribution layer rather than migrating their entire platform.

This is important. Many of you have a Shopify storefront with agentic commerce turned on by default. But Shopify Catalog lets brands that aren’t on Shopify yet upload their catalog as well. Shopify is essentially saying, “We’ll be the master product graph for AI commerce, and it doesn’t really matter where your store is built.”

The protocol war is about to be settled. The Universal Commerce Protocol, or UCP, is winning. I’ve written about it a few times, and a lot of other people are talking about it. It’s backed by some huge names: Walmart, Target, Etsy, American Express, MasterCard, Stripe, Visa, Microsoft, and more. There are thousands—not 30—thousands of Shopify merchants live and selling through it.

Even Meta has reportedly extended UCP into social commerce. They’re adopting it, and I think that’s an amazing development.

There is one conspicuously absent player, and it’s one most operators aren’t really pricing in yet: Amazon. That’s the elephant in the room.

Amazon is not part of UCP, and they’re not part of OpenAI’s protocol either. They’re betting entirely on Rufus inside their own walled garden. If you’re a brand doing meaningful Amazon volume, this is a distribution risk worth modeling, because I believe the AI discovery layer is consolidating around Shopify, not Amazon. That’s a strategic divergence that could really matter in the next 12 to 18 months.

So where do we go from here?

I always want this show to be a bit tactical, not just strategic. Here’s what I think you should do this quarter, depending on your maturity and complexity.

If you’re under $500K in annual revenue and early in your growth stage, my advice is pretty direct: don’t lose a single hour of engineering or operational time on agentic commerce strategy work this quarter. Instead, focus on what I call the boring fundamentals.

Get your product data clean and structured. Write your product titles and descriptions the way real humans search for them, in plain language. Fix your metafields. Ship clean schema markup on your product pages. This is exactly the kind of prep work that will make you discoverable in AI when AI really starts to matter for your category.

It’s also just good ecommerce hygiene that pays off whether or not agentic commerce becomes a major channel for you in the future.

If you’re at the next level, scaling from roughly $2 million to $20 million, the move is a bit different.

First, confirm that agentic storefronts are turned on inside your Shopify admin. They should be on by default for eligible stores, but double-check and verify that. Then audit your top 25 product titles for what people are calling Answer Engine Optimization, or AEO.

The test is simple: when a customer asks ChatGPT for the best item in your category under a specific price, does your product show up?

Start tracking AI referral traffic in your analytics. Don’t spend a dollar optimizing until you can see that you’re actually showing up. Once you see that traffic, then it makes sense to invest.

If you’re in the high-scaling or enterprise side, above $20 million, you’ve got a couple of priorities to figure out.

First, make sure your knowledge base content inside Shopify is tight, because I believe that’s where AI agents are pulling from when they describe your brand and products. Your brand voice, FAQs, and policies all matter now.

Second, if you have heavy Amazon exposure, this is a strategic conversation to have with your team this quarter. The AI distribution layer is not consolidating around Amazon, and that asymmetry is starting to widen.

Overall, my universal advice across every stage—early, mid-market, or enterprise—is this: treat AI as a discovery channel that drives traffic to your store and optimize for that reality for now.

Ignore anyone trying to sell you on autonomous AI checkout in chat. The major players just told you that’s not where agentic commerce is going.

Before we wrap, I want to make sure this really lands with one clear call to action.

Agentic commerce is not dead. It’s just splitting into a discovery layer that’s consolidating into AI, and a transaction layer that’s staying with you, the merchant. Brands that understand this cycle won’t be the ones chasing the next announcement that drops in two months or six months. They’ll be the ones who have built clean, structured, AI-readable product data while everyone else is distracted by the wrong story.

If you want to go back and hear where I started this conversation, check out episode 453 from April 1st. This episode is the update.

This whole topic maps directly to what I’m covering in the AI implementation chapter and the Convergence Advantage chapter of my upcoming book, *Thriving with Shopify*. If you want more of this thinking week to week, the place to get it is inside Fastlane Insider, my newsletter.

The link will be in the show notes below. It’s leading-edge thinking and very practical.

Thank you so much for being here today. I really do appreciate you. I hope this has been impactful for you, and we’ll see you in the next episode of eCommerce Fastlane.