The Maze: Product discovery used to begin with a keyword, a retailer search box or a category page. It increasingly begins with a conversation. In BCG’s global consumer surveys, the share of people who had used GenAI for brand or product search and recommendations rose from 32% in February 2025 to 36% in April 2026. That is only four percentage points. It is enough to change the first gate a brand must pass: not merely being found, but being judged useful by a system that compares options before the shopper reaches a product page.
GenAI research is already mainstream before it becomes shopping-specific. General information seeking was the largest use case in both waves, rising from 51% to 53%. That matters because product research does not live in a sealed commerce funnel. A shopper can move from “help me understand this problem” to “which product solves it” inside the same prompt. Brands built for one tidy query and one final click will miss the earlier moment when needs, constraints and alternatives are being assembled.
Brand and product research gained four points, from 32% to 36%. The source labels that movement an 11% relative increase. The safer management metric is the visible +4 percentage-point change: more than one in three respondents in the April 2026 wave had personally used GenAI for brand or product search and recommendations. This is not proof that AI made the purchase or picked the winner. It is proof that a meaningful share of consumers are inviting AI into the consideration set, where a recommendation can narrow the field before conventional retail media gets a chance to compete.
The other use cases show the same decision-assistance pattern. Travel planning rose from 19% to 23%, while health-care support rose from 15% to 18%. These are separate survey purposes, not parts of a 100% total, but they share a mechanism: people use GenAI when comparing complex options or turning a vague intention into a next step. Product discovery belongs in that family. The prize is not a prettier chatbot answer. It is a better ability to appear when the shopper asks a contextual question that keyword data would never fully capture.
Visibility is necessary; recommendation-worthiness is harder. BCG’s wider argument is that agentic systems can compare reviews, pricing, availability, service outcomes and other observable evidence alongside a brand’s own claims. That makes structured product data useful, but insufficient. A well-formed feed can get a product into the room. Delivery reliability, return friction, price integrity and customer sentiment influence whether it stays on the shortlist. In the agent-mediated journey, operations become part of marketing evidence.
Why it matters: The customer journey is gaining a second evaluator. Humans still choose, but AI increasingly helps frame the shortlist and tests the story a brand tells against the experience it delivers. Commerce leaders should measure which needs trigger AI research, whether their offer is cited or recommended, and which service or product facts cause the answer to favor a rival. The response is not to chase a single “AI ranking.” It is to align the promise, the product data and the operational reality before the question is asked.
Sources: BCG, “Agentic AI Is Redefining Marketing Growth”, Exhibit 3; BCG Global Consumer Radar Wave 3 (February 2025, n=7,286) and Wave 6 (April 2026, n=12,117).
