It’s Time to Fire the Chatbot
Summary
Build a smarter chatbot. Prove it works on 50 e-commerce sites. Job done, right?
Wrong.
Our AI chatbot stayed stuck in the dim corner of the online store, and few shoppers engaged.
Then we made one small shift—and everything started to change.
The Big Idea
In 2023, we launched a generative AI chatbot for e-commerce with high hopes.
What did we learn? Even the most advanced AI falls short when it’s stuck behind the complaints desk.
Here’s the inside story of why we gave our chatbot the pink slip— and why you should do the same.
1 – Start with the Problem to Solve
When generative AI hit the mainstream, we had a choice: Stick with what we knew—delivering best-in-class chat support in e-commerce—or take a harder, more meaningful turn: reimagine how people shop online.
We chose the latter.
Why? Because brands spend up to 20% of their revenue driving traffic—only to greet visitors with filters, grids, walls of text, and autoplay videos. No wonder 97% bounce without buying. Meanwhile, in-store conversion rates range from 20–40%.
That gap? That’s the opportunity.
We asked:
- What if online shopping didn’t feel like a research project?
- What if the website adapted to each consumer?
- Could we make the shopping experience faster, more engaging—even fun?
That's the experience we set out to build.
2 – Prove that AI Drives Conversions
In 2023, we launched one of the first generative AI chatbots at Cdiscount —supporting 80 million products, 2,500 categories, and 20 million monthly visitors.
The results were promising—24% conversion rates at Cdiscount. Generative AI chatbots could now answer pre-sales questions and drive conversions. They had become legitimate CRO tools.
By the end of the year, our AI chatbot was live on 50+ e-commerce sites.
3 – Make the Shift: Think Outside the Chatbox
But something still felt off. It didn’t feel like we had changed the game.
The problem? 📉 Fewer than 1% of shoppers clicked on the chatbot.
This wasn’t just our issue—it was systemic. AI chatbots were supposed to reinvent online shopping. But they all end up in the same place: a dim corner of the site, waiting to be clicked. Reactive. Isolated. Passive.
Even ours—technically impressive, capable of mimicking selling behavior and reducing hallucinations—was just… ignored.
We built an AI sales rep—and then confined it to a dim corner of the store.
It was time to think outside the chatbox.
4 – Study and Replicate the Bright Spots
We found inspiration in the “bright spots,” popularized by best-selling authors Chip and Dan Heath—those rare successes that show the way forward. Here’s one we’ll never forget.
Just before Christmas 2023, we met Adrien Delavallade, Head of Omnichannel CX at Nespresso.
Adrien didn’t want to reduce contacts—he wanted to boost shopper engagement. That was music to our ears.
We reviewed new AI engagement formats—ones that anticipated what shoppers wanted on product pages or search results, and dynamically personalized the experience.
Adrien and his team voted. They picked what we now call a conversation starter—context-aware, personalized AI that integrates natively into the shopping experience.
Nespresso's AI engagement format—iAdvize Conversation Starters
“I really like this one — it feels like it’s part of our site,” Adrien explained.
That was our aha moment: AI must blend into the shopping experience—and elevate it.
We ran A/B tests through January and couldn't believe the results:
📱 34x higher engagement on mobile
🖥 11x higher on desktop
Since then, this idea has scaled fast.
- Kendra Scott’s SVP Kamanashish Kundu saw a 9x engagement rate compared to the AI chatbot.
- Mademoiselle bio’s E-commerce Director Cédric Gourgeon said: “I was convinced AI would never beat our best advisors...In February, we hit 6x our pre-AI revenue.”
- This bright spot was replicated across 150+ e-commerce sites in the next year.
📈 The verdict? Engagement jumped 10x.
Turns out, Adrien was right.
5 – Fire the Chatbot
Consumers have voted with their clicks.
Fewer than 1% engage with the chatbot. That tells us where it belongs: on the contact or help page, handling what it does best—issue resolution.
But for the other 99%—the shoppers, the browsers, the buyers—you need something else.
Not an AI chatbot. Not a talking search bar. An eCommerce native shopping assistant or sidekick:
✅ Anticipates intent—steps in as an advisor, product expert, or support assistant
✅ Engages naturally—across shopping pages, not in a silo
✅ Advises like a top in-store associate—at e-commerce speed
It’s time to ⚠️ Fire the Chatbot— and recode your shopping experience.
Ready for 10x engagement and 6x revenue contribution?
Let’s talk. We’re already helping brands in fashion, beauty, jewelry, electronics, and home goods make the leap.
Talk to our team
