What is an AI shopping assistant?

A shopper-facing assistant operating as part of a retailer-controlled commerce experience: grounded in the retailer’s catalog, policies, and commercial context, and connected where relevant to inventory, cart, account, checkout, or service systems. It does not have to be built in-house — a technology vendor can supply the assistant while the retailer still controls the experience.

Do AI shopping assistants actually increase sales?

The largest disclosed result is Amazon’s: more than 300 million customers used its assistant in 2025, and the company attributes nearly $12 billion in incremental annualized sales to it. Metric definitions differ, though — some figures are management’s own incrementality claims and others are associations between use and purchase — so they should not be collapsed into a single benchmark.

Does an AI assistant replace onsite search?

No. Search remains one of the fastest interfaces in commerce when intent is explicit, and roughly half of shoppers on large general e-commerce sites prefer search while the other half prefer navigation. An assistant is most useful as an additional interface for situations where the shopper’s need is easier to describe than to encode as a product query.

How do I know the sales are incremental and not cannibalized?

By measuring against a comparable holdout rather than comparing assistant users with everyone who ignored it. Etsy’s A/B test is the cautionary case: a 28.3% lift in recommendation clicks produced no significant change in conversion or GMV, because the gain was offset by displaced organic search activity that no attributed metric would have revealed.

Yes. Leave your work email to read the full report right here — free.

AI Shopping Assistants for E-Commerce

The Business Case for AI Agents That Sell

About 30% of U.S. consumers now use generative AI to compare products and decide what to buy, and Amazon’s assistant reached more than 300 million customers in 2025. The conversational storefront has stopped being a demo.

Get the Free Report See What's Inside

Get a sneak peek of the report at

of U.S. consumers use generative AI for product comparisons and recommendations: ~30%
of U.S. consumers use generative AI for product comparisons and recommendations
in incremental annualized sales Amazon attributes to its shopping assistant: $12B
in incremental annualized sales Amazon attributes to its shopping assistant
the conversion Magalu’s WhatsApp assistant reached against its own app and website: 3×
the conversion Magalu’s WhatsApp assistant reached against its own app and website
of global online holiday sales were influenced by AI and agents: ~20%
of global online holiday sales were influenced by AI and agents

Search monetizes the intent a shopper can already express; the assistant works on everything they cannot.

Report Overview

What the retailer results show, where a conversation adds sales rather than moving the same order, and how to size a pilot that survives returns, cannibalization and operating cost.

Get the Free Report

01Foreword
02What we mean by an AI shopping assistant
03Executive summary
04The intent e-commerce can already monetize
05From product query to shopping brief
06What the deployments already show
07Seven sources of revenue, not just conversion
08Proactivity: who gets offered help
09Shopping tasks, not product categories
10Customer service as a sales channel
11The limits of the current evidence
12How to design the pilot
13The retailer’s assistant vs. general-purpose AI
14Commercial constraints, not just technical ones
15The end of the shopper’s translation burden

Key Findings for Your Team

Explore the findings most relevant to your team — each section is tailored to a different set of priorities.

E-Commerce & Merchandising
Where richer intent expression beats an already-functioning search box. Covers the shift from product query to shopping brief, multi-item missions, and why understanding a mission and executing it are separate design problems.
Customer Service & Operations
Service is a sales channel, not a cost centre. Covers revenue protection over deflection, the return flow as a product-discovery moment, and post-purchase context that creates legitimate cross-sell.
Growth & Analytics
The incrementality question, in full. Covers holdout design, the Etsy cannibalization result, first-principles pilot economics, and why assistant-attributed revenue should not be the primary KPI.
Retail Media & Marketing
Conversation is a new first-party intent surface. Covers Amazon’s Sponsored Products and Brand Prompts inside its assistant, proactive outreach, and where sponsored recommendations have to stay visibly distinct.

Where in your customer journey do shoppers know what they want but fail to say it in a form your store can monetize?

Bart TurczynskiHead of Growth, Tidio

Get the Free Report

About 30% of U.S. consumers now use generative AI to compare products and decide what to buy. What the retailer results show, where a conversation adds sales rather than moving the same order, and how to size a pilot that survives returns, cannibalization and operating cost.

Frequently Asked Questions

A shopper-facing assistant operating as part of a retailer-controlled commerce experience: grounded in the retailer’s catalog, policies, and commercial context, and connected where relevant to inventory, cart, account, checkout, or service systems. It does not have to be built in-house — a technology vendor can supply the assistant while the retailer still controls the experience.

Support agent working at a laptop next to the Lyro mascot

Ready to automate customer service with a quality agent?

Configure Lyro from your editor, or stand it up from our UI. Same agent, same skills, same evals — your choice of entry point.

Get started Talk to sales