The news: AI is inspiring consumers to change their purchase decisions and abandon traditional search channels, though its use remains moderately low.

Just 19% of US adults, or about 49.6 million people, start their retail product research with AI, per PYMNTS. And 39 million of that cohort have moved their retail research entirely to AI, leaving behind resources like Google Search, Amazon, and review sites.

Among those who are incorporating AI in their shopping journeys, 83% have made at least one change to their buying choices, including finding a better price (43%), choosing a different brand (27%), or buying a different product (26%).

What it means: A top placement on a search engine results page (SERP) like Google Search or Amazon answers won’t matter to the growing subset of shoppers who never see those results at all. And the rise of AI search opens up doors for challenger brands.

AI-driven switching behavior could make commoditized products, where price and specs drive purchases, more vulnerable. It could also challenge incumbents—10.8 million US adults who use AI for retail research have bought from a totally new brand, per PYMNTS.

An AI tool that’s used to compare products doesn’t focus on whether a retailer dominates shelves or paid search, which could flatten competitive advantages and make market share less important.

Recommendations for retailers: Those who wait for broader AI shopping adoption to invest in visibility may lose out on customers. Big-name brands need to prepare to compete with a wider market as AI search surfaces new challengers.

• Distinguish AI-referred traffic to gain insights on how pricing, product positioning, brand voice, and presence across third-party websites are attracting AI retail researchers.
• Audit AI assistants monthly to see how generative engine optimization (GEO) strategies are working and where they need model-by-model tweaking.
• Evaluate which product categories have the highest rates of AI-driven brand switching and optimize product data through clear descriptions and review sections to increase the likelihood of surfacing in answers.