The AI shopping blind spot: buyers who never reach your funnel

7 min readAshton
A shopper asking an AI assistant for a product while an unnamed brand's storefront sits outside the light, the AI shopping blind spot

Buyers increasingly ask an AI assistant to recommend a product before they open a search engine, and the assistant names two or three brands, not ten links. If yours is not one of them, you are not being outranked. You are not in the running.

This is happening at the top of the funnel, before a shopper lands on a brand's site, which is why retargeting cannot reach it. The available response is not a new ad channel. It is publishing product content specific and structured enough for an assistant to quote directly: exact ingredients, honest comparisons, real specifications, not persuasive copy.

Why are DTC buyers asking an AI assistant before they search Google?

A shopper comparing electrolyte powders or retinol serums used to open a search engine and scan a page of results. Now a growing share opens ChatGPT, Perplexity, or a phone's built-in assistant and asks it to just pick one, because a direct recommendation is faster than reading five product pages and guessing.

General adoption of AI chatbots keeps climbing: Pew Research found 49 percent of US adults now use one, up from 33 percent two years earlier. Referral traffic to retail sites from generative AI sources is growing even faster, off a much smaller base: Adobe Analytics logged a 1,300 percent year-over-year jump in the 2024 holiday season, and Similarweb found AI referral traffic to beauty sites up 312.5 percent year over year through May 2026. None of these figures say what share of a DTC shopper's research actually starts inside an assistant rather than a search bar.

What happens to a brand an AI assistant never names?

A search results page has ten slots and a scroll bar, so even a brand on page two gets some clicks. A conversational answer has room for two or three names before the reply stops being useful. Leave your brand out of that first answer, and a shopper who never scrolls past the reply never learns you exist.

This is a consideration-set problem, not a ranking problem. Ranking eleventh for a competitive term still gets a determined shopper to click through eventually. Never being named in an assistant's answer means you were never in the running at all. We have written before about why ranking and being cited are different problems, and product discovery through an assistant runs entirely on the second one.

Why can't retargeting catch this kind of shopper?

Retargeting depends on a pixel firing, which depends on a browser loading your page. A shopper who asks an assistant for the best magnesium supplement and clicks straight to checkout on the named brand's site never lands on yours. There is no visit to retarget, no cookie to drop, no audience to build.

The retargeting stack, the pixel, the abandoned-cart email, the dynamic product ad, assumes a shopper already reached your site and needs a reason to come back. It recovers interest. It does not create it. A decade of optimizing the bottom of the funnel does nothing if the top of that funnel has quietly moved somewhere the tracking cannot see.

Why is this traffic so hard to see in your own analytics?

Assistant referrals are still a small slice of total traffic for most sites, so a handful of sessions get lost in the noise of a busy dashboard. Many also arrive with no referrer, because an app or an in-chat browser does not always pass one, so analytics tools log the visit as direct traffic instead of crediting the assistant that sent it.

That gap is why this shift is easy to dismiss inside a single brand's own dashboard, even while it shows up clearly in aggregated data from providers who can see traffic across thousands of sites at once.

The defensible response: publish what a model can actually quote

An assistant can only quote what exists somewhere in clear, specific, structured language it can retrieve. That means exact formulations and dosages instead of "clinically proven," named comparisons against competitors instead of vague superiority claims, and third-party test results instead of marketing copy. Write the sentence you want quoted verbatim.

This is early, and the measurement needed to prove one comparison page earned one citation is still poor. No analytics provider can yet draw a clean line from a page to an assistant answer to a sale. That is a reason to start now and track carefully, not to wait for certainty that is not coming. It is the kind of work behind our growth systems for DTC brands.

Key Takeaways

  • An AI assistant answers with two or three named brands instead of a page of links, so a brand left out of that short list is not in the shopper's consideration set.
  • Retargeting, remarketing, and abandoned-cart flows only work after a shopper lands on a site, and a shopper who gets a direct answer inside an assistant may never land anywhere but the brand it named.
  • AI chatbot adoption has grown quickly, reaching 49 percent of US adults in 2026 per Pew Research, up from 33 percent in 2024, and referral traffic to retail sites from generative AI sources is growing even faster per Adobe Analytics.
  • Assistant referral sessions are undercounted in most analytics setups because they are a small share of traffic and often arrive without a referrer, so the shift can be invisible in one brand's dashboard even as it shows up in aggregated industry data.
  • The available defense is publishing specific, structured, comparison-ready content, such as exact formulations, direct competitor comparisons, and third-party test results, since this is a citation problem more than a ranking problem.

FAQ

Does this replace search engine optimization for a DTC brand?

No. Search still drives most organic commercial traffic for DTC brands today. This is an additional, earlier layer of visibility to manage, alongside SEO rather than instead of it.

How do I know if an AI assistant is recommending my brand?

Ask the assistants yourself, using the exact phrasing a buyer would type, such as "best magnesium glycinate for sleep" or "compare the top electrolyte brands." Run the same prompts across ChatGPT, Perplexity, Claude, and Gemini regularly, since each draws on different sources. There is no reliable automated dashboard for this yet, so manual, repeated checking is currently the honest method.

What does "structured content" actually mean for a supplement or skincare brand?

It means stating exact facts in plain sentences: milligrams per serving, the specific form of an active ingredient, what a third-party lab test found, and how your product compares to a named competitor on one attribute. It means avoiding vague claims like "premium quality," which give a model nothing concrete to quote.

Is it worth acting on this now, given how early and unmeasured it is?

Early does not mean it is safe to ignore. The underlying shift, buyers asking an assistant instead of searching, looks durable even though the tools to measure it are still immature. Start with changes that also help human readers and search rankings, since those carry no downside either way.

Who is behind Second Brain's view on this?

Second Brain is a growth consultancy that tracks which answer engines send DTC brands traffic and what they quote, following the sourcing rules in our editorial policy. You can read more about the people who run Second Brain.

Sources