ChatGPT to take a 4% cut for direct checkout: How brands should prepare

Shopify has confirmed that it will soon enable direct checkout through AI chatbots, including ChatGPT. The rollout will launch on January 26th.

Merchants whose products are purchased via ChatGPT’s checkout will pay OpenAI a 4% fee on those transactions. This is in addition to Shopify’s existing charges, such as payments processing and platform fees.

Shopify also confirmed that merchants will be included by default in AI checkout experiences that do not charge an additional fee. For now, transactions completed through AI experiences run by Google and Microsoft will not incur extra costs.

This is one of the clearest signals yet that agent-driven commerce is moving from experimentation into real economic infrastructure.

Why OpenAI is charging, and why now

The decision to take a 4% cut is not surprising when viewed through the lens of OpenAI’s incentives.

OpenAI operates at enormous scale and cost. Unlike Google, it does not own a global distribution monopoly in search or advertising. Monetising transactions is one of the most direct ways for OpenAI to demonstrate that AI agents can generate revenue, not just engagement.

Checkout is also strategically important. It allows OpenAI to sit closer to the moment of value creation, rather than acting purely as a discovery layer that hands traffic elsewhere.

In that sense, the fee is not just about commerce. It is about proving that conversational interfaces can replace parts of the traditional funnel and capture value accordingly.

Is 4% actually expensive?

At first glance, a 4% fee feels meaningful, especially for merchants operating on thin margins. But relative comparisons tell a different story.

On Amazon, referral fees often exceed 10%, sometimes significantly more depending on category. Those fees also come with deep platform dependency, limited brand control, and restricted access to customer relationships.

Compared with paid acquisition, the contrast is even sharper. Meta, Google Ads, and other performance channels routinely require double-digit percentages of revenue to remain competitive, especially in crowded categories.

What makes AI checkout particularly interesting is the likelihood that many of these sales will be incremental. Agent-driven purchases are often triggered by highly specific intent. For example, a user might ask for “waterproof black running shoes in size 9 for winter commuting”. That level of specificity does not always translate well to traditional ads or browsing behaviour.

In those cases, the AI agent is not cannibalising an existing sale. It is unlocking demand that might otherwise have gone unserved.

Seen through that lens, the 4% fee looks like a fairly reasonable acquisition cost from an incredibly powerful discovery channel.

Google’s free approach reveals a deeper divide

Google’s decision not to charge additional fees for AI checkout, at least for now, is a strategic one.

Google already monetises commerce indirectly through search, shopping ads, and data. Making checkout frictionless strengthens its core business. Immediate transaction-level monetisation is not essential.

OpenAI is in a different position. It needs to demonstrate a credible, scalable business model that extends beyond subscriptions and enterprise licensing. Charging for checkout is one of the fastest ways to do that.

This divergence matters for merchants. It suggests that AI commerce will not be priced uniformly. Different agents will optimise for different incentives, and brands will need to understand how those incentives shape recommendation behaviour.

How Shopify merchants should prepare for agent-driven checkout

1. Product data becomes the competitive surface

Agents rely on structured, machine-readable information to evaluate suitability, compare options, and make decisions on behalf of users. If your data is incomplete or inconsistent, your product simply becomes harder to choose.

Merchants should prioritise:

• Accurate and complete product schema markup

• Clean, well-maintained product feeds

• Consistent data across websites, Merchant Center, and third-party platforms

• Rich attributes covering variants, compatibility, materials, and use cases

Google is already expanding Merchant Center with dozens of new attributes. Brands that invest the effort to populate these fields will be easier for agents to understand and trust.

2. Brand authority now matters to machines

Agents do not just evaluate products. They evaluate merchants.

They draw on a range of sources to assess brand credibility, including owned content, third-party mentions, and structured brand data. If your brand is inconsistently represented or poorly documented, agents may avoid recommending it altogether.

Merchants should focus on:

• Publishing consistent, high-quality content across owned channels

• Implementing organisational schema markup

• Ensuring brand information is accurate across major platforms

• Building earned mentions in sources AI systems commonly reference

Tools like Azoma can help identify which sources agents rely on, where your brand appears, and where it does not.

3. Consistency beats cleverness in fulfilment

Pricing accuracy, stock availability, delivery promises, and returns policies all play a role in agent decision-making.

Agents are risk-averse by design. They prefer merchants who provide stable, up-to-date information across every surface. Even small inconsistencies can reduce the likelihood of being selected.

Merchants should ensure that fulfilment logic and commercial terms are synchronised everywhere they appear.

4. Sentiment becomes a ranking signal

Agents do not read reviews the way humans do. They analyse patterns.

This means sentiment at a SKU and category level matters, not just overall star ratings. Repeated complaints about sizing, delivery, or support are likely to influence recommendations, even if average scores remain high.

Merchants should:

• Track sentiment at a granular level

• Invest in post-purchase experience and support

• Encourage genuine reviews and user-generated content

• Reinforce the attributes they want agents to associate with their products

Much like PageRank rewarded consistent credibility, agentic systems reward sustained, verifiable customer satisfaction.

5. Add Context to reduce friction

Agents need clear context to match products to user intent.

Merchants should explicitly define who a product is for, when it should be used, and when it should not. Compatibility, limitations, and edge cases should be stated clearly rather than implied.

The goal is to remove ambiguity. The less an agent has to infer, the more confident it can be in selecting your product.

6. Measurement needs to evolve

Traditional metrics such as clicks and impressions capture less value in a world where agents transact directly.

Merchants should begin tracking:

• How often their brand appears in AI responses

• Inclusion and exclusion from agent recommendations

• Visibility across different AI platforms and regions

Agent visibility should be treated as a top-of-funnel metric in its own right.

➡️ Azoma gives your brand visibility into how you appear across AI systems. It shows when and where your brand and products are being surfaced, cited, or missed by large language models and agentic interfaces.

The bigger picture

The 4% fee is not just about pricing. It signals a shift in how digital commerce is starting to work. AI tools are moving beyond simple discovery and into the mechanics of buying and selling.

For merchants, the change is straightforward. Winning will rely less on bidding for clicks and more on being easy to understand, reliable, and clearly relevant when customers are looking for something specific.

Brands that adapt early will not only unlock incremental sales. They will also influence how this new layer of commerce develops and which merchants it favours over time.

Azoma is already helping some of the world’s largest brands, including Mars, L'Oréal, and Colgate, navigate this shift towards agent-driven commerce. If you want to understand how AI systems see, evaluate, and recommend your products, get in touch with Azoma today.