Shopify says AI shoppers reward specific products. Here’s what CPG brands should fix in their product data

Shopify reported third-quarter results on November 3. The headline was the usual one: gross merchandise volume of $119.1 billion, up 29.5% from a year ago, and revenue up 32%. The part a CPG brand should read twice came later in the call, when Harley Finkelstein described what AI shopping on Shopify actually looks like. AI-referred orders grew 3.3x year over year. And the shoppers arriving from AI tools were not buying the obvious stuff.

“AI channels are surfacing highly specific products that fit specific customers with a specific need,” he said. Here’s what Shopify disclosed, what a new survey says shoppers do after reading an AI answer, and what Google changed on brand searches the same week.

What did Shopify say about AI shopping in Q3 2026?#

Finkelstein was careful to say the volume is still small, but “adoption is growing.” He gave a handful of multiples, all comparing AI channels with something else. AI traffic to Shopify stores grew 2.6x year over year, and orders grew faster than traffic, which means AI-referred visits are converting better over time, not worse.

AI-referred orders, year over year3.3x

AI-referred traffic, year over year2.6x

AI shoppers sent straight to product pages, vs. organic search3x

Catalog-powered searches converting, vs. general AI search2x

AI shoppers converting (“64% more often”)1.64x

Source: Shopify Q3 2026 earnings call, Nov 3, 2026 (transcript via The Motley Fool)

Two of those numbers matter most for product teams. AI channels send shoppers straight to a product page at 3x the rate of organic search, so the product page is the landing page now, not the homepage or a category grid. And searches powered by Shopify Catalog convert at 2x the rate of general AI search, which Finkelstein contrasted with “what people call scraping.” Catalog turns more than a billion product pages into structured data, translating details like size, color and material into values an agent can compare.

Put plainly: when an agent reads clean attributes instead of guessing from page copy, twice as many of those searches end in a sale.

Why are AI shoppers buying outside the top categories?#

The most striking number on the call: 74% of AI-attributed purchases in Q3 came from outside Shopify’s top 100 categories. CFO Jeff Hoffmeister repeated it, saying three quarters of AI-referred sales “yet again” came from outside the top 100. So this is not a one-quarter fluke.

• Outside Shopify’s top 100 categories74%
• Inside the top 10026%

Share of AI-attributed purchases, per Harley Finkelstein on the call. The CFO described it as three quarters of AI-referred sales, “yet again.” Shopify did not publish a category list.

Source: Shopify Q3 2026 earnings call, Nov 3, 2026 (transcript via The Motley Fool)

This is what happens when shoppers describe a need instead of typing a product name. Nobody asks ChatGPT for “shampoo.” They ask for a sulfate-free shampoo for color-treated curly hair that is safe for a sensitive scalp. Every clause in that prompt is a filter. A product that states those facts in its data survives the filter. A product that says “gentle, salon-quality care for every hair type” does not, because the agent can’t prove it matches.

For CPG, the long tail is not only niche brands. It is the specific variant of a big brand: the fragrance-free pod, the 12-count multipack, the nut-free bar. Your top seller may be fine. The variant that matches a precise prompt is where AI demand is showing up, and it is usually the one with the thinnest product data.

Do shoppers still click after reading an AI answer?#

Mostly, yes. A Fractl study published in Search Engine Land on November 2 surveyed 1,007 US adults and paired it with clickstream data. When an AI answer appears at the top of results, 56% said they read it and then click a link to double-check. Only 16% read the answer and stop.

Read it, then click to double-check56%

Scroll past it22%

Read it and stop16%

Search again3%

Don’t notice it3%

Source: Kelsey Libert (Fractl), “The quiet click,” Search Engine Land, Nov 2, 2026 (survey of 1,007 US adults)

Seventy-one percent said they want a real source before a final decision or purchase. Of the clicks that landed on content, 31% went to product pages and 20% to reviews. The top reasons people clicked were a trusted expert (24%), original data (23%) and customer reviews (17%). Among clicks that left the original search page, Amazon drew 13% and Reddit 7%. One caution: shopping was a lower-stakes topic in the survey, with 27% preferring a real source there versus 72% for health. Shoppers verify, but they verify less for a $6 purchase.

Read that next to Shopify’s numbers. The AI answer shortlists, and then a shopper or an agent checks the product page, the reviews and the retailer listing. If those disagree with each other, you lose at the check.

What changed on Google brand searches this week?#

Google appears to be moving AI Overviews for brand searches out of the main results and into the right-hand panel, where the knowledge panel used to sit. Search Engine Roundtable reported it on November 2, after SEOs including David McSweeney spotted it on October 31. Google has not announced it. One of the examples was Coca-Cola, where the panel summarized the brand under the label “Carbonated soft drinks.”

1. Sep 29, 2026AhrefsAI Overviews appeared on 82.91% of branded results in its tracking, per PPC Land. Source
2. Oct 31, 2026Search communitySEOs spot brand AI Overviews moving to the right-hand panel. Source
3. Nov 1, 2026GoogleProduct panels in Search gain Add to cart and Buy now in the US, per PPC Land. Source
4. Nov 2, 2026FractlSurvey: 56% of US adults read an AI answer, then click to double-check. Source
5. Nov 3, 2026ShopifyAI-referred orders up 3.3x; 74% of AI-attributed purchases outside its top 100 categories. Source

A brand panel written by a model is only as good as what the model reads about you. In one of the reported examples, Zillow’s panel gave conflicting founding years. For a CPG brand, the equivalent risk is a summary that gets your category, ingredients or claims wrong in the most visible spot on your own brand search. Check yours this week.

What should CPG brands fix in their product data?#

Shopify’s data says agents reward specifics. Fractl’s says shoppers check the specifics before buying. The work is the same for both:

1. State the filters as facts. Count, size, scent, form and claims like gluten-free or fragrance-free should be in structured attributes and in the first lines of the description, not buried in a bullet at the end. That is the input Catalog turns into comparable values.
2. Give every variant its own page and data. The 74% lives in variants and niches. A multipack that shares a description with the single unit is invisible to a precise prompt.
3. Make the product page win the double-check. AI tools send shoppers straight there. Put the facts a shopper verifies (ingredients, dimensions, certifications, reviews) above the fold.
4. Match your retailer listings. If Amazon says 48 loads and your site says 50, an agent or a shopper has a reason to doubt both. We covered how ChatGPT ranks stores for a product in our merchant ranking breakdown.
5. Turn on the agent paths you already have. OpenAI says Shopify merchants reach ChatGPT shopping through Shopify Catalog with no extra work. Per Shopify’s help center, direct checkout in Google AI Mode and Gemini is on by default for eligible stores selling to the US, Canada or Australia, with no extra fees beyond payment processing. It does not support subscription-only products or bundles, which matters for replenishment brands.

Here is a quick way to test the first point. Paste a product description and the claims a shopper would filter on. The test checks whether each claim shows up in the first 40 words, the same test we used in our Walmart agents post.

If the claims only appear after the brand story, move them up. The story can stay. It just shouldn’t be the first thing an agent reads.

Our take#

Shopify gave the clearest public data yet that AI shopping is a specificity game. Generic hero copy was built for shelf presence and brand search. AI channels reward the product that can prove it fits one person’s exact need, then survive a shopper’s double-check on the product page. For more context, see our Xenalytics replica on Shopify’s GMV growth and the glossary entry on agentic commerce.

Want to know which of your SKUs show up for specific prompts and which drop out? Run a free audit or book a demo. Or see how SearchPanel scores AI visibility per product.