Gema Calderón Sayoux

vender productos en ia

How Much Does Your Ecommerce Store Make Thanks to AI? ChatGPT, SEO and Sales in 2026

If your answer is “no idea”, this article is for you. And if your answer is “zero, because I don’t sell inside ChatGPT”, it’s for you too. Because that statement is no longer entirely true.

In 2025 OpenAI began introducing instant checkout inside ChatGPT and, as of today, certain eligible products and merchants can let the user complete the purchase without leaving the chat. For the rest, ChatGPT can display the product, compare options and send the user directly to the store.

But even that falls short of explaining what’s happening. The important question for an ecommerce store is no longer simply:

“Can I sell inside ChatGPT?”

The question is:

When someone asks an AI to recommend what to buy, does my product appear among the options, or does my brand not even enter the conversation?

AI is becoming a new layer for product discovery, comparison and recommendation. And that layer can influence a purchase, even when the final transaction still takes place on your own website.

Ecommerce keeps growing. And so does the battle for visibility.

Ecommerce in Spain generated €29,296 million in the third quarter of 2025, 19.3% more than in the same quarter of the previous year. In that period there were also more than 508 million transactions, up 14.8% year over year. This is data from the CNMC.

In other words: we’re not talking about a small market or a technology looking for a problem that doesn’t exist.

We’re talking about an ecommerce market that keeps growing and a new way of searching for products that is starting to enter that buying journey.

For years we’ve measured digital visibility with questions like:

  • What position do I appear in on Google?
  • How many impressions do I get?
  • How many clicks do I receive?
  • How much organic traffic do I generate?

All of these are still important metrics. But now there’s another one:

What happens when the user doesn’t search for ten results, but asks an AI to recommend three?

Traffic from AI is growing at an extraordinary speed

The available data doesn’t yet allow us to say that AI has replaced Google as an acquisition channel. But it does allow us to say that the growth of traffic coming from AI tools is extraordinary.

Adobe Analytics analyzed more than a trillion visits to US retail sites during the 2025 holiday season. Between November and December, traffic to retail sites from generative AI tools increased 693.4% year over year. In November the increase was 769% and in December 673%.

According to Adobe, during that season visits coming from AI converted 31% better than the rest of the traffic and revenue per visit from AI increased by 254 % compared with the comparison period. On Thanksgiving the conversion difference reached 54% and on Black Friday 38%.

In addition, visitors coming from AI showed deeper behavior:

  • less likely to leave immediately;
  • more interaction with the site;
  • and signs of higher purchase intent.

But careful: AI is not Google yet

There’s an enormous difference between saying: “AI traffic is growing 693%.” and saying: “AI is already the main sales channel for ecommerce.”

The second statement is not supported by the available data. In fact, one of the most comprehensive academic studies published so far shows a much more nuanced picture.

Maximilian Kaiser and Christian Schulze analyzed first-party Google Analytics data for 973 ecommerce sites, with about $20 billion in combined revenue. The study observed more than 50,000 transactions coming from ChatGPT and compared them with about 164 million transactions from traditional channels.

One year after ChatGPT’s launch, organic traffic from LLMs:

  • converted better than paid social;
  • but still converted worse than traditional channels;
  • and represented less than 0.2% of traffic in the sample.

In addition, ChatGPT accounted for more than 90% of the observed sessions from LLM platforms; Perplexity, Gemini, Copilot and the rest had much smaller volumes.

So, how much money is AI really generating?

There is still no public, homogeneous database that lets us state: ChatGPT generates X% of all ecommerce sales.

We also can’t directly compare:

Google → X% of sales
ChatGPT → X% of sales
Perplexity → X% of sales

On a global scale with a single reliable source.

The available studies use different methodologies. For example, Salesforce estimated that during the 2025 holiday season AI and agents influenced approximately $262 billion, equivalent to 20% of the global online retail sales it analyzed. Salesforce also found that users who reached retailers from AI-powered search channels, such as ChatGPT and Perplexity, converted nine times more than those coming from social referrals.

From searching for products on Google to asking AI for recommendations

For years the typical behavior was:

User → Google → results → store → product → purchase

Now another journey appears:

User → AI → conversation → comparison → recommendation → product → store → purchase

And there’s even a third scenario:

User → AI → product → checkout inside the AI experience itself

OpenAI already allows, in the United States, for certain eligible products and merchants, completing payment directly from ChatGPT through Instant Checkout. For other products, the user continues to the merchant’s website.

Therefore, ecommerce stores need to start thinking in three levels of visibility.

What does a product page need to be readable by an AI?

There’s no single “optimization for ChatGPT” and no magic tag that guarantees appearing in a recommendation. What you can do is make sure the information that search engines, assistants and recommendation systems need is complete, consistent, up to date and accessible.

Implement the schema for Productcorrectly, including where applicable:

  • name;
  • brand;
  • SKU;
  • identifiers;
  • image;
  • description;
  • variants;
  • price;
  • currency;
  • availability;
  • offer conditions.

And keep that data in sync with what actually appears on the page. A Product schema that says €99 while the product page shows €119 helps no one.

Structured product data can become a fundamental layer of interoperability between your catalog and different discovery and commerce systems.

A good feed should avoid:

  • duplicate products;
  • inconsistent SKUs;
  • incorrect categories;
  • outdated prices;
  • incorrect stock;
  • empty attributes;
  • poorly related variants;
  • nonexistent images.

The more structured your catalog is, the less work a machine has to do to understand it.

Don’t hide important information inside an image, a PDF or a catalog. 

If a product has:

  • material;
  • dimensions;
  • weight;
  • compatibility;
  • size;
  • color;
  • composition;
  • ingredients;
  • certifications;
  • battery life;
  • power;
  • warranty;

that information should be available in a clear, structured way

Shopping systems need up-to-date information. OpenAI’s own documentation warns that prices, availability, discounts and conditions can change and that there may be some delay in updating the information.

That’s why:

product + price + stock + URL

must be aligned.

When someone asks: “Which of these three options can I buy and receive before Friday?”, availability and logistics can completely change the answer.

That’s why it’s worth making these easily accessible:

  • shipping costs;
  • delivery areas;
  • estimated times;
  • return policy;
  • return costs;
  • warranty;
  • restrictions.

AI can use that information to help compare options, but the user must be able to verify the final conditions with the merchant.

And how do I know if I’m already selling thanks to AI?

This is the question every ecommerce store should ask. In GA4, review the sources and mediums generating traffic and purchases. Look for referrals such as:

  • chatgpt.com
  • perplexity.ai
  • gemini.google.com
  • copilot.microsoft.com

And measure, at a minimum:

Metric What it tells you
Sessions
How much traffic does it generate?
Purchases
How many purchases does it generate?
Revenue
How much money does it generate?
Revenue/session
How much value does each visit produce?

The checklist to prepare your ecommerce store for AI search

  • Up-to-date feed
  • No duplicate products
  • Categories correctly assigned
  • Prices in sync
  • Stock in sync
  • Canonical URLs
  • Valid images
  • Variants correctly represented
  • Product Schema correctly implemented
  • Consistent price and currency
  • Up-to-date stock
  • Variants correctly related
  • Consistent SKU and identifiers
  • Brand correctly defined
  • Complete description
  • Technical attributes visible
  • Quality images
  • Compatibility information where applicable
  • Shipping clearly explained
  • Accessible returns
  • Visible warranties
  • Clear commercial terms
  • Contact information
  • Company information
  • Accessible policies
  • GA4 ecommerce correctly configured
  • purchase working
  • Revenue correctly recorded
  • Source/medium correctly captured
  • AI referrals monitored
  • Conversion rate by channel
  • Revenue by channel
  • Revenue per session
  • Attribution beyond last click when possible

AI doesn’t replace Google yet, since the traffic it generates is still small compared with the major traditional channels.

But it’s growing quickly; it can show high purchase intent and it’s already entering the buying process itself. Adobe observed a 693.4% growth in AI traffic to retail during the 2025 holiday season; the academic study by Kaiser and Schulze shows the channel is still small in volume, but with different economic behavior from other channels; and OpenAI already allows purchases inside ChatGPT for certain merchants and products.

Sources and methodology

Leave a Reply

Your email address will not be published. Required fields are marked *