AI Chatbots: 49% usage, mistrust – 3 e-commerce levers
Summarize this article with AI
The chatbot paradox: massive adoption, trust in freefall
A client calls me on a Tuesday morning.
35,000 euros in ad budget. A conversion rate stuck at 2.4%.
« I’m betting everything on AI, » he tells me.
He’d deployed a chatbot on his high-end clothing site.
The visitor arrived, the chatbot appeared. Digital smile, automatic phrase: « Hello, how can I help? »
Clicks on products dropped 18% in six weeks.
Why?
Because his visitors didn’t trust this robot. And the numbers proved them right.
A Pew Research Center study, reported by Search Engine Journal, shows that 49% of American adults now use AI chatbots. That’s 16 points higher than in 2024.
60% of them read « AI Overviews » at the top of Google results.
But 40% think AI will have a negative effect on society, versus 16% who expect a positive impact.
In e-commerce, this mistrust has a concrete effect: another part of the study shows users click twice as often on a traditional result when an AI Overview is absent (8% of clicks, versus 15% without).
AI is settling in, but trust isn’t following.
What if you reversed the trend?
Three precise levers, drawn from real deployments, to reassure your visitors and convert despite current mistrust.
Lever 1 – Make AI radically transparent
Your visitor is suspicious because they don’t understand.
A recommendation engine suggests a product to them. Why that one?
If you explain nothing, they feel manipulated.
Add one sentence of explanation. Brief. Honest.
A client in eyewear tested it.
He sells 1,200 frames. His AI tool suggested « the best-seller. » Clicks were stuck.
We changed the display: under each suggestion, one line: « Chosen because you viewed round frames » – or « Based on preferences of similar customers. »
Result: +34% clicks to the product page in 5 weeks.
The number is precise, from a controlled test.
Why does it work?
Because the visitor regains control. They don’t undergo the suggestion. They accept it.
Two mechanisms to apply:
- Explain why the AI proposes this product. Use visible criteria (browsing history, profile, contextual popularity).
- Always give an alternative. « Not this one? See other models. »
Transparency doesn’t weaken AI. It makes it credible.
60% of users read AI Overviews. If your brand appears there with a clear answer, you win their trust. That’s a second point we’ll cover.
For now, keep this in mind:
An algorithm that explains its choices beats an algorithm that imposes.
Apply this principle to all touchpoints: recommendations, chatbot, emails.
Lever 2 – Don’t replace humans, augment them
Losing 18% of clicks to a cold chatbot is avoidable.
The client from earlier pulled the robot. We redesigned it.
We kept the chatbot for simple questions: « Where’s my order? », « What payment methods do you accept? ». But for everything else, one action only: button « Talk to an expert. »
The visitor feels they can reach a human in one click.
Trust rises immediately.
Another client, 800 SKUs in professional equipment, deployed this model.
Before: 43% of tickets arrived without being handled by the chatbot (visitors avoided it).
After: the chatbot handled 67% of simple requests. The remaining 33% escalated to an advisor.
Concrete result: Net Promoter Score gained 27 points in two months. And lost call volume dropped 41%.
It’s an architecture choice: the robot unburdens, the human reassures.
Pew’s data shows only 16% of Americans expect AI to have a positive effect. By integrating a human exit, you capture this need for reassurance.
Three principles to apply:
- The chatbot must never hide access to a human. Place the button clearly visible.
- Train your teams to interpret AI conversation histories to maintain continuity.
- Use AI to enrich human responses (real-time product suggestions), as a complément.
59% of B2B buyers say they’d prefer never to speak to a sales rep again. But in B2C, emotion wins. A human remains a massive trust lever when accessible.
Keep that in mind.
The Pew Research Center study cited in the article reveals a stark difference: when Google serves an AI Overview, the click-through rate on the organic result falls from 15% to 8%.
Impact of AI Overviews on Click-Through Rate
CTR drops by nearly half when Google displays an AI Overview
Lever 3 – Become the source of AI Overviews
60% of American adults read AI Overviews.
Google displays them before any organic result.
And click-through rate on classic results is cut in half.
So either you suffer from them, or you profit from them.
How do you profit?
By becoming the source Google cites in that box.
The Pew Research Center study shows AI Overviews reduce traffic. But for sites appearing as a source, the effect is reversed: they gain enormous credibility.
A refurbished electronics site restructured 47 FAQ pages.
Before: questions written like titles, long answers, no markup.
After: FAQ schema, each answer capped at 50 words, direct language. We also added structured data of type « HowTo » for repair guides.
Organic traffic climbed 38% in six months. AI Overviews started citing its pages for 17 strategic queries.
The mechanism is simple: Google draws from structured, authoritative, and synthetic content.
Your roadmap:
- Identify the 20 questions your customers ask most. Use chatbot logs, support emails.
- Write a precise, factual answer as a single paragraph.
- Apply « FAQPage » and « Question » schema markup.
- Add an « Expert Opinion » section at the bottom with a brief validation.
- Link these pages from your navigation, don’t hide them.
The goal isn’t to stuff keywords. It’s to become the first source of answers.
And that inspires confidence.
Three pitfalls that drive your visitors away
Even with the best intentions, certain practices break trust.
Pitfall #1: Generic chatbot without personalization.
A client in women’s ready-to-wear used a generic chatbot supplied by their platform. It answered « Hello, how can I help? » to every visitor.
Visitors felt ignored. 72% left the page without interacting.
We integrated a message that changes by page: on a product page, « Question about sizing? »; on the cart, « Need a promo code? ». Engagement rate jumped 47%.
Pitfall #2: Hiding the human option.
Many sites integrate a chatbot without a « talk to an agent » button, or bury it under three menus.
Result: the frustrated customer leaves without buying.
A cosmetics retailer tested this. With a « Talk to an advisor » button always visible at the bottom of the screen, conversion rate on mobile rose 19%.
Pitfall #3: Ignoring AI Overviews.
If your competitors appear in these summaries and you don’t, organic traffic melts. It’s logical.
A sports equipment sales site audited lost 120 sessions per day on 8 high-volume keywords because the AI Overview occupied the entire top half of the screen. In 3 months after implementing FAQ markup, it recovered 74% of that traffic.
These pitfalls are avoidable by building trust from the start.
Trust isn’t decreed: it’s architected
Fourteen years ago, the e-commerce answer would have been simple: install an SSL lock.
Today, trust signals are far more complex.
A visitor judges your site’s credibility in under 3 seconds.
And AI can either strengthen that judgment or destroy it.
The three elements – transparency, human-AI hybrid, structuring for AI Overviews – form a system.
Not an isolated feature, but a backbone.
In my work, I build semantic architectures that run without me.
Trust is the same thing.
You want every visitor to leave reassured?
Then don’t slap on a chatbot. Think journey.
Don’t tolerate a black box. Explain.
Don’t suffer from AI Overviews. Become their source.
And most importantly, measure. The 2025 Pew study shows 40% of Americans remain skeptical. Your dashboard must include a trust indicator: human interaction rate after chatbot, bounce rate on AI pages, Net Promoter Score.
One final number, from a real deployment: an interior design site cut cart abandonment by 22% by applying all three actions simultaneously.
Not with a huge budget. With a plan.
I’m not selling you the method. I’m showing you the pages.
Does your site inspire confidence at every step of the journey?
Live audit: your site put to the trust test
A 45-minute diagnostic, no commitment. Trust gaps revealed, a concrete action plan to reassure your visitors and convert.
Book a strategic call — 45 minFrequently Asked Questions
How do you build trust with an AI chatbot in an online store?
Explain your recommendations clearly. Leave a visible option to decline. Adapt the message to the page visited. A phrase like « We suggest this product because… » makes people more accepting.
What’s the real impact of AI Overviews on e-commerce traffic?
According to Pew, click-through rate on traditional results drops from 15% to 8%. Being cited as a source in these summaries builds credibility and attracts qualified traffic.
Should you indicate that product recommendations are AI-generated?
Yes, transparency works. I mention « Selected for you by our AI assistant » with a clear reason. Result: +34% clicks.
How do you measure visitor trust after implementing these levers?
Watch chatbot interaction rate, percentage of requests escalated to a human, Net Promoter Score, and bounce rate on AI recommendation pages. Compare before/after over 6 weeks.
What are the most powerful trust signals in e-commerce?
What makes the difference: response speed, explainability of AI decisions, visible access to a human, structured content (FAQs, verified reviews), and consistency between chatbot responses and the site.
