E-commerce

E-commerce: Shoppers ask AI before they buy. Do the language models mention your store?

Hans Audun Sørensen·3 October 2026·7 min read

E-commerce: Shoppers ask AI before they buy. Do the language models mention your store?

The share of Norwegians who use AI when shopping online went from 22 to 32 percent in less than a year. That means a large part of your customers have asked ChatGPT, Gemini or Google for advice before they ever see your store. Black Friday falls on 27 November this year, six weeks from now. If you start now, you can get your product data into a shape the language models can read before the campaigns begin.

How many shoppers use AI?

Prisjakt has measured this twice. In June 2025, 22 percent of Norwegians said they used AI tools when shopping. By March 2026 the share had reached 32 percent, according to a nationally representative survey of 1,004 people carried out by Opinion for Prisjakt. The most common uses are comparing products and looking up product information.

The same survey shows trust lagging behind. Only 20 percent say they trust information from AI tools, 70 percent fear more fake product reviews, and 66 percent worry about more fraudulent online stores. Keep that in mind for the rest of this piece: shoppers use AI to find and narrow down, but they want to check for themselves before they pay.

There are no public figures yet for how much traffic AI assistants send to Norwegian online stores. That is a good reason to measure it yourself. Point 9 below shows how.

How AI assistants recommend products and stores

ChatGPT, Gemini, Copilot, Perplexity and Google AI Mode answer questions like "which winter jacket works for cycling to work, under 300 euros" with a handful of products and a short explanation. To make it into that answer, your store has to be one of the sources the assistant finds and trusts.

Broadly, the assistants draw on three places:

Search indexes. ChatGPT, Copilot and Perplexity usually search the web when they answer shopping questions, and Gemini and AI Mode build on Google's index. If your product pages are easy for a machine to read, they are more likely to be used.
Product feeds. Google uses data from Merchant Center, including in AI Mode. OpenAI accepts product catalogues, but so far only from stores in the US. A feed delivers price, stock status and product details in a format the model does not have to interpret.
What others say about you. Reviews, comparison sites, forums and product tests. If your store or product is mentioned there, the assistant is more likely to mention it too.

Exactly how each assistant weighs these sources is not documented, and it keeps changing. What you control is the quality of your own data and how easy it is to retrieve.

What is available in Norway today

Google launched AI Mode in Norway in October 2025, with support for Norwegian. It builds on Google's regular index and Merchant Center, so the work you do for Google Search and Shopping counts here too.

ChatGPT is a more mixed picture. The shopping feature, where stores supply their product catalogue directly to ChatGPT, is according to OpenAI available only in the US. Buying directly inside ChatGPT never reached Norway, and OpenAI has since scaled it back so that the purchase is completed on the store's own site. What has arrived here is advertising. Since August 2026, ChatGPT ads have run in 31 European countries, Norway included. The ads are labelled and kept separate from the answers.

For a Norwegian online store, this means visibility in ChatGPT currently depends on what the assistant finds on the web and what others write about you. The sale still happens in your own checkout, on your terms and with your customer data. The rest of the buying journey has to hold up when a customer arrives from a chat window with high expectations and low trust.

Nine things to do before Black Friday and Christmas 2026

Check what the assistants say about you today. Write down ten questions a customer might ask in your category and put them to ChatGPT, Gemini, Copilot, Perplexity and Google AI Mode. Note which stores and products come up, and whether the information about you is correct. Repeat it in early November so you have something to compare against.
Clean up your product feed in Google Merchant Center. Google requires a unique ID, title, description, link, image, availability and price, among other things, and recommends GTIN and brand where available. Go through the errors in Merchant Center and fix them. Mismatches between the feed and the product page are a common reason products get disapproved.
Consider ChatGPT as an advertising channel. The ads are new in Norway, and nobody has much experience with them yet. If shopping questions to AI assistants matter in your category, a small test before Black Friday gives you early experience. Keep an eye on chatgpt.com/merchants as well, where OpenAI announces when the shopping feature expands to more countries.
Add structured product data. Mark up product pages with schema.org Product and Offer: name, brand, GTIN, price, currency, availability and aggregate rating. Google says that structured data on the page combined with a Merchant Center feed gives it the best basis for understanding and verifying your product data. Other assistants reading the page get the same facts without having to guess.
Write product copy that answers questions. Assistants get questions like "is this suitable for" and "what's the difference between". Copy that just repeats the manufacturer's marketing gives them little to work with. Say who the product is for, dimensions and materials, what it is not suited for, and how it differs from the closest alternative in your range.
Collect real reviews. Reviews are among the things assistants cite when they explain a recommendation. Ask customers for a review after delivery, show both good and bad ones, and mark them up with structured data. When 70 percent of Norwegians fear fake reviews, being open about how you collect them is a competitive advantage.
Keep prices and stock status up to date. An assistant that recommends a product at the wrong price, or one that is out of stock, sends you a disappointed customer. Make sure your feed updates often enough to keep up with price changes during Black Friday week, and test it before the campaigns start.
Make checkout fast and trustworthy. A customer arriving from an AI recommendation has often made up their mind but does not necessarily trust your store yet. Show shipping costs and delivery times early, have a clear returns policy and contact details, and offer payment methods people recognise. An unnecessary step in checkout costs more when the customer has already chosen the product somewhere else.
Measure AI traffic separately. Set up a dedicated channel group in your analytics tool for traffic from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. According to OpenAI, ChatGPT adds the parameter utm_source=chatgpt.com to its links, which makes that traffic easy to isolate. Traffic from Google AI Mode does not show up as its own channel. It is counted as organic search.

What is realistic in six weeks

Most stores can handle points 1, 2, 7 and 9 on their own before November. Structured data and new product copy take longer, so start with the categories that sell most over Christmas. Whatever you get done, the customer coming from a chat window should find what the assistant promised: the right product, the right price and a checkout that works.

At Frontkom I work with e-commerce and help stores with product data, integrations and buying journeys. I have also written about how AI is changing the way we shop (in Norwegian).

Sources

Transparency note: This article is based on the author's own analysis and experience. AI has been used as an editorial aid for language and structure.

E-commerce

Platform vs. store: What's the right e-commerce strategy for your business?

7 min read