ChatGPT isn't sending you traffic. It's sending you decisions.

When ecommerce teams look at their analytics and see a small number of sessions attributed to AI assistants, the reaction is usually the same. The numbers are too small to act on. There are bigger priorities. Come back to this in six months.

The problem with that logic is that it measures AI search the same way you measure every other channel, and AI search does not work like every other channel.

The conversion gap nobody is talking about

A shopper who arrives from a ChatGPT recommendation has already done the research. They asked a question, received a curated answer that named your product, and chose to click through. By the time they land on your product page, the consideration phase is largely complete. They are not browsing, they are confirming.

That is a fundamentally different visitor from someone who clicked a Google result or a paid ad. And it shows in the numbers.

Shopify's own data shows AI-driven orders on the platform grew 15x year-over-year in 2025. Research consistently shows AI-referred sessions convert at three to four times the rate of paid prospecting traffic, with higher average order values and lower bounce rates.

At a £2 to £4 CPC on branded or category terms, 100 ChatGPT sessions converting at 4x your paid rate is the equivalent of £600 to £1,600 of paid media value, earned, with no media spend behind it. At 500 sessions a month, that is a meaningful number on any growth team's spreadsheet.

The question is not whether the volume is high enough to act on. The question is whether your brand is showing up at all.

Why your analytics are understating the impact

Most of the commercial influence of AI search never appears in your AI referral data at all.

Shoppers increasingly use ChatGPT as a discovery and research tool, not a navigation tool. They get a shortlist, a set of product attributes to look for, a brand name to remember. Then they come back later through a different channel: a branded Google search, a direct visit, an email click, a paid retargeting ad. In your analytics, that session shows up as organic, direct, or paid and the ChatGPT conversation that started the whole thing is invisible.

This means your AI-referred sessions are a floor, not a ceiling. The actual commercial contribution of AI search to your revenue is higher than what your attribution model shows, and it grows every time a shopper uses an AI assistant to shortlist products in your category.

Why low AI traffic is a diagnostic, not a verdict

If your brand is seeing very low AI-referred sessions, it is one of three things. Here is how to tell which:

You are not appearing in AI answers at all. Open ChatGPT and ask three to five queries that your best customer would ask when searching for your product category. If your brand does not appear in the responses, your products are not in the recommendation set. Shoppers asking those questions are being sent to competitors instead.

You are appearing, but not getting the click. If your brand name shows up in AI responses but sessions are still low, the recommendation is not confident or specific enough to drive action. AI assistants recommend with conviction when they have clear, structured product evidence to work with. Vague descriptions, missing attributes, and incomplete structured data all reduce the specificity of the recommendation and the likelihood of a click-through.

You are appearing and getting the click, but losing the attribution. If your branded search and direct traffic are healthy but AI referrals look low, this is likely an attribution gap rather than a visibility gap. The AI is influencing decisions that are being credited elsewhere. This is actually the best of the three problems to have, because it means your products are already in the recommendation set, and fixing the attribution gives you a clearer picture of the channel you are already benefiting from.

Each of these is diagnosable. But you cannot diagnose what you cannot see, which is why the starting point is always product-level visibility data, not brand-level traffic numbers.

The compounding effect most teams are sleeping on

AI models form associations over time. The brands appearing consistently in answers to relevant queries are the ones the model learns to reference. Citations lead to more citations and early presence creates a pattern that compounds as the channel grows.

The brands building AI visibility now are not just capturing today's AI-referred sessions. They are establishing the citation authority that will make them the default recommendation when AI search becomes a much larger share of product discovery. Shopify's data already shows the trajectory: 15x order growth in a single year.

Brands that wait until AI search traffic is large enough to justify investment will be starting from behind in a channel where the leaders have already established authority, at a cost of entry that will only increase as competition for AI recommendations intensifies.

The number worth finding out this week

Product-level visibility is where the commercial opportunity actually sits. Brand-level visibility tells you very little. Knowing that your brand appears somewhere in AI answers is not the same as knowing that your hero SKUs are being recommended for their core queries.

Glara runs your product catalogue against the queries that matter for your category and tells you specifically which products are appearing in AI answers, which are invisible, and what attributes are driving the gap. For a head of ecommerce or growth, that is the difference between knowing you have an AI visibility problem and knowing exactly which products to fix first and what the revenue implication is.

Do you know how many of your top twenty SKUs are currently recommended by ChatGPT for their core query? If not, that is the number worth finding out this week.

Book a demo and Glara will show you exactly which products are visible, which are not, and what it would take to close the gap.