OpenAI recently introduced Instant Checkout, a feature born out of their partnership with Stripe. The idea is simple. A customer asks for product recommendations. ChatGPT shows options. If the merchant supports Instant Checkout, the customer can buy immediately, inside the chat itself. No tabs. No website visit. No browsing journey. No second thoughts. Just search to purchase in a single window. So if you asked someone where they got their cool jacket from, they could potentially say it's from ChatGPT instead of the actual seller.
All this sounds great in theory, but will it work in real life?
"It's the latest hype cycle for ChatGPT that's simply unable to sustain the growth that got it so much investment in the first place."
- A skeptical user on Reddit.
Not all hype turns into history. Despite ChatGPT’s arrival, we haven’t seen a single D2C brand become an overnight sensation because of it. And with the 1.4 trillion the company has poured into expanding its business, you'd think that there'd be solid outcomes and results plastered everywhere. And yet, all we see in terms of evidence is... nothing much, actually.
Given the less than stellar results around AI shopping and ChatGPT, it's better to realistically analyze whether you should hop on Instant Checkout or if you should just let things be. Let's first look at how it actually works, and then whether it's worth all the hype.
According to OpenAI, Instant Checkout is "the next step in agentic commerce." Imagine you're a customer who just asked ChatGPT to search for “best housewarming gifts under $100”. It pulls products from across the web.
If the seller's website supports Instant Checkout, you can hit buy, fill in your details, and the order is placed, without ever leaving the chat. The customers don't have to pay a dime to OpenAI to use this feature, but you do. If merchants want to have it enabled for their business, they'll have to pay a fee on completed purchases.
On their website, OpenAI mentioned that Instant Checkout items are not prioritized over others in product results. But then they added that if someone asked ChatGPT to rank sellers on the same product, the model would consider factors like availability, price, quality, whether a merchant is the primary seller, and also whether Instant Checkout is enabled. This means that the feature won't be the sole deciding factor, but it will be one of the variables considered.
To understand the scale of what this means, consider the reach. By December 2025, ChatGPT had surpassed 810 million weekly active users. That is nearly 10 percent of the world’s adult population interacting inside a single interface.
For merchants, this raises important questions like: when the platform owns the interaction, the discovery, and the checkout, where does that leave you? Should you be panicking about it ruining your customer reach, traffic and attribution? And are people really going to use AI shopping features? The quick answer is no, not yet.
Let’s be honest about what people actually use ChatGPT for. Around 70–80% of sessions are for writing, obtaining information, or seeking advice. Only 2.1% involve products you can actually buy. Interestingly, self-expression has more volume than shopping
Clearly, that’s not a “ready to buy” crowd. It’s a “think, plan, draft” crowd.
And think about the actual moment of purchase. Until now, it’s always happened elsewhere, because ChatGPT didn’t have a checkout button. Now it does.
That changes the rails, but not necessarily the shopping behavior. Just because the buy button moved inside the chat doesn’t mean shoppers will skip the steps they trust most. Because who pays without comparing prices, checking reviews, scrolling through pictures, or even without asking their friends?
Shopping isn’t just about utility. It’s visual. It’s emotional. It’s comparative. Look at your own perspective when you're the consumer. People (including you) tend to buy when pictures persuade us, when reviews reduce the risk, and when browsing sparks the unexpected.
Remember those funky looking Labubu dolls? The same collectibles that crossed 8 billion USD in sales in 2025. These sales were not driven by convenience. They were driven by emotion, scarcity, community demand, and the fear of missing out. The fact that they were out of stock in multiple places only made them more desirable. Convenience did not create the demand; context did.
This is also why customers still choose options like cash on delivery in many markets. Not because it is faster, but because it feels safer. It adds a layer of reassurance in an uncertain transaction.
A single-item, text-driven checkout flow removes many of those reassurance layers.
Remember when Voice commerce, Messenger chatbots, AR mirrors all launched with billion-dollar projections? They all fizzled.
All worked technically. All failed behaviorally.
Merchants on reddit are saying the same thing,
"This is the equivalent to asking Alexa to shop for you, when was the last time you purchased anything through it? This is not how shopping works, the buyer’s journey is way longer than blindly giving ChatGPT a prompt."
So, is it persuasive? Not really.
Let’s move from theory to numbers.
Ahrefs studied the traffic makeup of 35,000 websites, and the AI referral results were disappointing, to say the least.
AI chatbots send just 0.1% of all referral traffic. Not 10%. Not 1%. A tenth of a percent. For context, that’s less than email referrals, and it’s nowhere near the reach of search.
Zoom in further, and the picture sharpens:
Which means that for the rest 99.9% businesses, the share is vanishingly small. Almost invisible.
By the time you filter AI referrals (0.1% of referrals) into the slice of total traffic (referrals usually make up 10% or less of a site’s mix), you’re left with something close to 0.01% of total traffic.
And then there’s the volatility. In July 2025, ChatGPT’s outbound clicks collapsed by 52% in a single month after OpenAI reweighted how it cites source pushing more visibility to Reddit and Wikipedia while suppressing long-tail sites.
It was a platform-level change in citation logic.
So, for an average e-commerce merchant the reality is:
In case if we are hoping that Instant Checkout might be a new channel, the math just doesn’t add up, at least not yet.
Even if Instant Checkout grows, merchants face a bigger structural risk: platform bias.
Generative AI tools aren’t built to be neutral traffic distributors. Their incentive is to keep users inside the chat and eventually monetize that flow. We’ve seen the same playbook with Google's AI overviews causing zero-click attribution, Meta penalizing links in feeds, and Amazon, with its self-preferencing paid placements.
ChatGPT is trending the same way. For example, Sam Altman, OpenAI's founder, used to be vehemently against ads on the platform. Now, he seems to have changed his mind.
Some operators already notice discovery blind spots:
“We have one of the biggest retro gaming store in our country but ChatGPT didn't recommend us. To my surprise it was under the impression that we are not a retro store. After a few follow-up question as to why that is, we had to rework our homepage and overall communication.”
Others warn of platform bias. One seller noted:
“You’ll just increasingly see preference for its feed-generated products.”
Meanwhile, other sellers also noted:
“Once Instant Checkout is rolled out, ChatGPT won’t search the web for products anymore but its own index of product feeds.”
Both signal the same concern: this isn’t neutral distribution. It’s a platform incentive to prioritize its own product feed, effectively competing with merchants for visibility.
But the real challenge lies in the business model. As one agency leader put it:
“There’s no option yet to retarget cart or product-page visitors.”
And without retargeting hooks, loyalty flows, or multi-item carts, Instant Checkout risks flattening the very mechanics that drive AOV and repeat purchase.
And that is only for today. OpenAI has already said:
“Today, Instant Checkout supports single-item purchases. Next, we’ll add multi-item carts and expand merchants and regions.”
Instant Checkout represents a larger shift for AI models and their race to dominate AI shopping. We can't yet determine what it means for merchants because changing buyer behavior requires more than just another convenience-oriented tech product. This is because ecommerce has never been majorly limited by the ability to complete a transaction. Trust has always played a bigger role.
And trust is built through signals like fleshed out product pages, realistic reviews, unbiased comparisons, brand familiarity, and visual confirmation. These are small moments that add up and reassure a customer that they are making the right decision. Instant Checkout removes friction, but it also removes many of those signals.
And history has shown that removing friction does not always create adoption; customers still gravitate toward experiences that gave them more context, not less. The same pattern may play out here.
The volume is not there yet. The traffic is not evenly distributed. The platform incentives are still evolving. And most importantly, customer behavior has not fundamentally shifted toward making high-trust purchases inside a conversation alone.
So maybe, for now, the only rational plan for us is to wait and watch. Or perhaps even ignore it until the behavior, not just the rails, shows signs of real adoption.
Don’t give in to the hype just yet. Some people/agencies will blow this up as the next big game, but you can sense the reality and see the playground for what it is, not what they’re making it out to be.
Because in e-commerce, behavior always beats hype. And right now, the behavior just isn’t buying inside the AI chatbot.
.jpg)
.jpg)
.jpg)