AI shopping: how people choose products and services with AI — and why they trust its recommendations
Jump to
Trust shows up as behavior: 68% say they’re willing to act on GenAI recommendations, and 58% prefer GenAI recommendations over traditional search. [S2]
GenAI is already a measurable commerce referral channel: during the 2024 holiday period, clicks from GenAI sources to U.S. retail sites rose +1300%, and on Cyber Monday +1950%. [S1]
People use AI for shopping decisions, not just content: research (55%), recommendations (47%), and finding deals (43%). [S1]
A meaningful share is already replacing search with GenAI: 24% used GenAI to find a product instead of a search engine; among ages 18–34 it’s 41%. [S3]
Not long ago, “shopping research” meant Google → marketplaces → 20 tabs → decision fatigue. Today, it increasingly looks like a single conversation: “Recommend the best options for my needs” — and a GenAI assistant returns a shortlist, with reasoning.
This is the new reality of ai for shopping: one conversation replaces a dozen tabs, and the shortlist becomes the shelf space that matters.
This shift matters not because “AI is trending,” but because the decision format has changed. People don’t browse endlessly. They accept an AI-generated summary, click one or two links, and move on. For brands, that means AI responses are becoming the new decision surface.
Below is an analysis of what 2024–2025 data tells us about choosing products and services with GenAI, where trust comes from, and what it means for brand AI visibility (which is what we focus on at Dabudai).
What changed in shopping: from “search results” to “one AI answer”
The key shift isn’t simply that people “trust AI.” It’s that GenAI optimizes the most painful part of shopping: information overload.
Historically, consumers had to:
compare specs,
read reviews,
figure out which criteria mattered.
Now they delegate that work to AI:
“Explain the differences,”
“Pick 3 options for my budget and use case,”
“What’s best for my constraints?”
This is what artificial intelligence shopping looks like in practice: the buyer delegates comparison and criteria-building to a model — then chooses from a compressed shortlist.
That creates a new attention bottleneck: the brands that make it into the AI shortlist win consideration
Market signals: GenAI is already driving warm traffic to retailers
Adobe Analytics captured a tangible surge in referral traffic from GenAI interfaces to U.S. retail websites (tracked via clicks from GenAI experiences). During Nov 1 – Dec 31, 2024, GenAI-driven traffic rose +1300% YoY, and on Cyber Monday it was +1950% YoY. [S1]
What’s even more interesting is how this audience behaves once it lands:
+8% higher engagement,
+12% more pages per visit,
–23% lower bounce rate compared to non-AI sources. [S1]
Analytically, that suggests GenAI helps users clarify criteria before clicking, which reduces “browsing noise” on-site. For brands, that means: making the AI answer is increasingly equivalent to entering the buyer’s consideration set.
How people use AI in shopping: research, recommendations, and deal-hunting
In Adobe’s survey (5,000 U.S. respondents), 39% said they have already used GenAI for online shopping, and 53% plan to do so “this year.”
The rise of ai in online shopping is not about novelty — it’s about compressing the entire research → recommendation → purchase flow into a faster decision cycle.
Top use cases:
research (55%),
product recommendations (47%),
finding deals (43%). [S1]
This matters because it’s not “AI for fun.” It’s AI as a decision assistant, especially at the discovery and consideration stages — before a user ever lands on a product page.
Trust that converts: people say they’re willing to act on GenAI recommendations
“Trust” can be fuzzy. In commerce, the practical version is: does it change what people do?
In other words, ai in shopping only matters when it changes behavior — clicks, shortlists, and purchases.
Capgemini’s research (12,000 consumers across 12 countries) reports:
68% are willing to act on GenAI recommendations, and
58% prefer GenAI recommendations over traditional search engines. [S2]
That’s the most practical definition of trust in shopping: a readiness to treat the recommendation as a basis for action.
GenAI is already replacing search for part of the market
Bazaarvoice’s Shopper Experience Index 2025 suggests GenAI is becoming a true discovery interface:
24% have used GenAI to find a product instead of a search engine,
among ages 18–34: 41%,
55% trust GenAI tools and “shopping agents” for at least some shopping tasks (ages 18–34: 75%). [S3]
From a brand perspective, the implication is blunt: if the customer’s first touch is an AI response, your SEO page may never get a chance unless you appear in that answer.
“AI recommended it — I bought it”: evidence that recommendations lead to purchases
Akeneo’s survey (Dynata, U.S.) connects recommendations to purchases:
32% reported completing a purchase based on an AI recommendation,
and 84% of those reported a positive experience.
75% noticed AI recommendations or chatbots during online shopping,
44% of those engaged with them. [S4]
This reinforces the main point: trust isn’t just sentiment — it shows up as purchases and satisfaction
Services: travel and financial services are scaling fast
For services, GenAI hits a sweet spot: complex constraints, lots of options, and repetitive comparisons.
Travel: Adobe reports GenAI traffic to travel/leisure/hospitality sites grew +1700% (Feb 2025 vs Jul 2024). They also report 29% have used GenAI for travel tasks, and 84% of those said it improved their experience. [S1]
Oliver Wyman (U.S. and Canada) reports 41% have used GenAI for travel inspiration or itinerary planning, rising from 34% in Aug 2023. And 58% say they’re likely to use it again (among recent users: 82%). [S7]
Financial services: Adobe reports GenAI traffic to banking sites grew +1200% (Feb 2025 vs Jul 2024), and 27% reported using GenAI for banking/financial needs. [S1]
Why people trust AI shopping recommendations
The “trust mechanism” is often misunderstood. It’s not that consumers suddenly believe models are perfect. It’s that GenAI reduces the friction of choosing.
1. Decision relief: AI compresses complexity
GenAI turns an overwhelming category into a manageable shortlist. That reduces decision fatigue — especially in spec-heavy categories and services.
2. Format signals expertise
Clear structure (“best for X,” “trade-offs,” “alternatives”) reads like expertise. Even without understanding “how it works,” consumers often treat the output as authoritative.
3. Trust grows with familiarity
Ipsos shows a clear adoption curve: among the most frequent GenAI users, 81% say they would trust AI recommendations; among less frequent users, trust is materially lower. [S5]
4. Transparency is the condition for scale
Salesforce reports consumers want guardrails: nearly 75% want to know they’re interacting with an AI agent; 44% are more likely to use it when the agent explains its logic; 45% when there’s a clear escalation path to a human. [S6]
Friends advice vs AI recommendation
Overall, people still trust humans more than LLMs for recommendations: Ipsos reports 89% trust recommendations from friends/family versus 38% for LLM recommendations (and 51% for AI recommendations on an e-commerce site based on purchase history). [S5]
But AI can outperform “people advice” in certain shopping contexts. For example, UserTesting reports 64% trust AI tools as much as or more than friends/family for gift advice (Gen Z: 76%). [S8]
What this means for brands: AI answers are becoming the new shelf space
In classic SEO, you fought for a position in search results. In AI-mediated shopping, you fight for inclusion in the shortlist a model produces “for this user’s constraints.”
In shopping with ai, visibility becomes the probability of being included in the answer — and conversion becomes winning the click among one or two links.
That changes the playbook:
visibility becomes the probability of inclusion in the answer,
trust becomes the quality of evidence signals (structured facts, credible mentions, authentic reviews/UGC),
conversion becomes winning the click from a compressed shortlist — not competing across ten blue links.
We’re also moving toward “agentic shopping,” where AI agents make choices on a user’s behalf. Early research on AI shopping agents suggests agents respond to attributes like price/ratings/review volume and platform presentation signals. [S9] Even if this is still early, the direction is clear: machine-readable credibility is becoming a competitive advantage
What we think about this at Dabudai
Dabudai is focused on helping brands manage how they appear in AI answers:
where the brand already shows up (and in what categories),
where it’s missing from the shortlist,
what sources/signals are shaping the model’s view,
how to reduce incorrect or outdated brand descriptions.
This isn’t “prompt magic.” It’s about strengthening what models pick up as evidence: consistent facts, structured content, credible mentions, and trustworthy UGC.
FAQ
Are people really using GenAI instead of search to find products?
Yes: 24% say they’ve used GenAI to find a product instead of a search engine (41% among ages 18–34). [S3]
Do GenAI recommendations actually lead to purchases?
Yes: 32% reported completing a purchase based on an AI recommendation, and 84% reported a positive experience. [S4]
How strong is willingness to act on GenAI recommendations?
Capgemini reports 68% are willing to act, and 58% prefer GenAI recommendations over traditional search. [S2]
Sources:
[S1] Adobe Analytics — GenAI traffic to retail/travel/banking + survey & engagement metrics
[S2] Capgemini Research Institute — What matters to today’s consumer 2025 (68% act; 58% prefer GenAI recs over search)
https://www.capgemini.com/insights/research-library/what-matters-to-todays-consumer-2025/
[S3] Bazaarvoice — Shopper Experience Index 2025 (GenAI replacing search; trust by age group)
[S4] Akeneo (Dynata survey) — AI-powered shopping tools: purchases, satisfaction, adoption, transparency concerns
[S5] Ipsos — Trust in recommendations: friends/family vs LLM vs e-commerce AI; trust among frequent users
[S6] Salesforce — AI Connected Customer research: disclosure, explainability, human escalation
https://www.salesforce.com/news/stories/ai-customer-research/
[S7] Oliver Wyman — GenAI in leisure travel planning and inspiration
[S8] UserTesting — AI as a holiday shopping companion (gift advice trust vs friends/family)
[S9] “What Is Your AI Agent Buying?” (agentic e-commerce behavior research)
