How it works
Both tools take the form of a conversation on the storefront, but they serve different moments of the shopper journey.
| Chatbot | AI Shopping Assistant | |
|---|---|---|
| Moment in the journey | After the purchase, for support | Before the purchase, while choosing |
| Understanding requests | Predefined scripts. Newer chatbots built on an LLM (Large Language Model) handle open-ended questions better | Open-ended questions, in the shopper's own words |
| Engagement | Waits for the shopper to open it | Suggested questions on the page being viewed, when the shopper hesitates |
| Catalog and stock access | Limited or none | Connected to the catalog and stock |
| Recommendations and cart | Sends the shopper to a page of the site | Shows products in the conversation, add to cart without leaving it |
| Questions it can't answer | Fallback message, form or handoff, depending on the tool | Hands over to an agent or creates a ticket in the support tool |
| Upkeep | Every new question needs a new script | Runs on the catalog and the brand's content, and flags gaps found in real conversations |
| Metric tracked | Tickets deflected | Revenue generated, measurable with an A/B test |
Why it matters in e-commerce
Many storefronts have installed a chatbot and find that very few shoppers use it. The problem lies in the tool's role: it waits for a question instead of helping the shopper choose.
An AI Shopping Assistant engages shoppers on the pages where they hesitate, with questions tailored to the product they are viewing. According to iAdvize data, it drives 10x more engagement than a traditional AI chatbot.
A support chatbot is enough when the need stops at after-sales: order tracking, returns, account questions. Once the storefront sells products that need advice (size, compatibility, use), the question comes up before the purchase, and that's where a chatbot doesn't help.
Further reading: It’s Time to Fire the Chatbot
Real-life example: Payne Glasses
On a sports equipment storefront, a chatbot waits in a corner of the screen. A shopper is torn between two pairs of trail running shoes and does not think to open it. With an AI Shopping Assistant, a suggested question appears on the product page: “Which shoe for muddy trails?”. She clicks, gets a comparison of the two models and adds the right one to her cart.
At Payne Glasses, many customers, often older, struggled with filters to find the right frames, and the small support team only worked business hours. With the iAdvize AI Shopping Assistant on product and listing pages, engagement is five times higher than with a traditional chatbot.
Common confusion
A common belief is that adding generative AI to a chatbot turns it into an AI Shopping Assistant. Generative AI improves how questions are understood, but a chatbot remains a support tool if it is not connected to the catalog, does not recommend products and is not measured on revenue.
Three questions settle it: does the tool see products and stock in real time? Can it add a product to the cart inside the conversation? Do you measure its effect on sales, not just on ticket volume?