Most consumers use AI, but few fully trust it
AI is now part of how people find products and make decisions. A new global study from Klaviyo, based on nearly 8,000 consumers, shows the split: 60% use AI at least weekly, but only 13% completely trust it. That gap is where the marketing work lives. People treat AI as an input, not an authority.
AI nudges purchases faster than trust grows
The study found 41% bought a product AI recommended in the past six months. Another 27% discovered a product via AI, then researched before buying. AI is acting like a discovery layer and first touchpoint. Consumers use it to narrow options, then verify before they commit.
Four AI personas you need to plan for
- AI Enthusiasts (≈26%): High usage and higher trust. 89% used AI while shopping in the past six months; 43% bought multiple products they hadn't known about before AI recommended them.
- AI Evaluators: Frequent users, but cautious. They compare and research with AI, then validate before acting. Together with Enthusiasts, they account for nearly 70% of consumers.
- AI Skeptics: Understand and occasionally use AI, but remain wary of AI in marketing and brand interactions.
- AI Holdouts (≈21%): Rarely use AI for shopping and prefer human guidance.
The real split isn't users vs. non-users. It's how much they trust the output.
Power users spot weak AI content instantly
Here's the catch: the people most fluent with AI are also the quickest critics. Among Enthusiasts, 40% notice low-quality or generic AI-generated marketing content multiple times per week. As usage grows, so does the ability to detect bland prompts, filler copy, and cookie-cutter "personalization." If execution slips, your most engaged audience bounces first.
Prompts are longer and more human
Search behavior is shifting. 78% of consumers add emotional or personal context to prompts at least some of the time, and 30% now use eight or more words. People ask complex, conversational questions. Your content and product data need to answer them cleanly and directly.
What to do now
- Design for AI discovery: Keep product data clean and complete (attributes, specs, price, availability, reviews). Publish clear FAQs and side-by-side comparisons to reduce decision friction. Use structured data where applicable so information is machine-readable. See Google's guidance on structured data for implementation details: Structured Data guide.
- Raise the quality bar: Set brand voice rules, provide high-quality examples, and require human edits. Fact-check claims, link to evidence, and avoid thin, generic copy. If a template can write it, your best customers can spot it.
- Build verification into the flow: Cite sources, surface reviews and ratings, and link directly to specs, shipping, and returns. Price and inventory should be live. Make it easy to jump to a human fast.
- Tune by persona:
- Enthusiasts: faster paths and multi-step workflows in chat.
- Evaluators: more evidence, comparisons, and links to verify.
- Skeptics/Holdouts: prominent human options (phone, chat), plain-language explanations.
- Optimize prompts and replies: Encourage natural-language questions. Offer short, scannable answers with one clear next step and an option to "show sources."
- Measure and iterate: Track AI-assisted sessions, acceptance of recommendations, verification clicks (to reviews/specs), time to human handoff, and post-purchase satisfaction. Use these signals to improve content and logic.
Guardrails that earn trust
- Be transparent: Tell people when AI is used and what data it relies on. Keep the explanation short.
- Control hallucinations: Answer only from approved content and provide fallbacks when confidence is low.
- Respect consent: Use first-party data with clear permission and easy opt-outs. Don't surprise people.
- Throttle automation: Cap frequency. Stacked triggers feel like spam and erode confidence fast.
- Audit for bias and risk: Align your process with reputable frameworks like the NIST AI Risk Management Framework.
KPIs built for the trust gap
- AI-assisted discovery rate: % of sessions or leads that begin from an AI touchpoint.
- Recommendation acceptance: Click-through or add-to-cart on AI suggestions.
- Evidence-seeking behavior: Clicks to reviews, FAQs, and specs after an AI interaction.
- Time to human handoff: How quickly people escalate from AI to a person.
- Complaint rate on AI content: Qualitative flags and support tickets mentioning AI.
- Repeat purchase from AI paths: Retention signals for AI-influenced buyers.
Upskill your team
Turn these insights into practice with focused training and playbooks. If you need a structured path for managers and ICs, start with the AI Learning Path for Marketing Managers.
Bottom line
AI use is mainstream, trust is not. Show up where discovery happens, but win by earning confidence: better data, stronger content, clear evidence, and respectful automation. The brands that help people verify will convert more, churn less, and build durable preference.
The full "Klaviyo AI Persona Research" report is available from Klaviyo (registration required).
Your membership also unlocks:
