How Consumers Buy in Google AI Mode: 6 Findings from 185 Purchase Tasks

About the Study

Kevin Indig, Garret French (Citation Labs), and Eric Van Buskirk (Clickstream Solutions) conducted a usability study with 48 participants completing 185 purchase tasks. Categories covered: televisions, laptops, washer/dryer sets, and car insurance. The key: a within-subjects A/B design where each participant used both AI Mode and traditional Google search.

1. 88% of Users Took the AI's Shortlist Outright

In classic search, 56% of participants built their own shortlist from multiple sources. In AI Mode, only 8 out of 147 codeable tasks produced a genuinely self-built shortlist.

64% of AI Mode users clicked nothing at all. They read the AI's text, scrolled through product cards, and named their finalists. The comparison phase wasn't just shortened. For most participants, it simply didn't happen.

2. The AI's Top Pick Becomes the User's Top Pick

74% of participants chose the item ranked first in the AI's response as their top pick. The mean rank of the final choice was 1.35. Only 10% chose something ranked third or lower.

26% of participants overrode rank order, but the driver was brand recognition. 81% of this group still chose from within the AI's candidate set.

3. The AI's Words Become the Trust Signal

AI framing (37%) and brand recognition (34%) were the top two trust drivers in AI Mode. In classic search, the dominant trust mechanism was multi-source convergence: participants built confidence by checking whether multiple independent sources agreed. That behavior was almost absent in AI Mode (5%).

"Travelers and USAA actually tell me how much, whereas State Farm and GEICO give percentages. Just knowing the exact amount makes me want to pick Travelers or USAA right off the bat."

4. If You're Not in the List, You Don't Exist

For laptops, three brands captured 93% of all AI Mode final choices. In classic search, the distribution was broader.

Two distinct problems emerged:

• Brands that never appeared in the AI's output were never considered. The AI decided who made the list, not the buyer.
• Brands that appeared but lacked recognition weren't seriously considered. Erie Insurance showed up in results but was eliminated by multiple participants purely on name recognition.

Notably, participants didn't feel constrained by the narrower set. Narrowness frustration appeared in 15% of AI Mode tasks vs. 11% of classic search tasks.

5. Users Leave to Buy, Not to Research

23% of AI Mode tasks involved an external site visit (vs. 67% in classic search). The critical difference: AI Mode users left to confirm a price or spec for a candidate they'd already accepted. Classic search users left to discover candidates.

Reddit appeared in 19% of standard search tasks but only twice across all 149 AI Mode sessions.

6. Three Strategic Levers

1. Visibility at the model layer: If AI Mode doesn't surface your brand, you have a visibility problem at the model layer. Query your own category the way a buyer would and document which brands appear regularly.
2. Framing: Brands cited with concrete attributes (specific price, model, use case) held stronger positions than generically described brands. The content on your site affects how specifically the AI describes you.
3. Structured pricing data: Where shopping panels showed explicit prices, 85% of participants understood pricing clearly. Where they didn't, confusion and overconfidence filled the gap. Schema markup and Merchant Center feeds are the most direct lever.

What This Means for Your Business

The study confirms a trend we at TYS have been communicating for months: AI visibility is no longer an optional channel. It's the new front door for purchase decisions. Our 7-layer framework addresses exactly these three levers: semantic data markup (Schema.org, llms.txt), structured product data, and context-specific content framing.

The full original study by Kevin Indig, Citation Labs, and Clickstream Solutions is available on Search Engine Journal.