Key statistics
- Customers are about 3 times more likely to use ChatGPT, Gemini or Copilot than a company's own chatbot to resolve a customer service issue, and use of company chatbots has been flat since 2022. Gartner
- 87% of customers say companies using generative AI for customer service must give them a way to reach a human agent. Gartner
- Nearly 1 in 5 consumers who used AI for customer service saw no benefit, a failure rate almost four times higher than for AI use in general. Qualtrics XM Institute
- Shoppers used retailers' own AI and agents for customer service 126% more during the 2025 holiday rush than in the two months before. Salesforce
A good AI chatbot can answer "Where's my order?" at 2am, help a shopper pick the right size and process a return without anyone on your team lifting a finger. A bad one traps customers in loops, invents policies and makes a simple question feel like a battle.
The technology has improved enormously in the past two years. Modern AI chatbots can understand natural questions, draw on your product catalog and policies, and increasingly take actions such as editing orders or issuing refunds.
Below, we'll cover what AI chatbots do for online stores, what the latest data says about how customers feel about them, how the main platforms price them and how to set one up well.
What is an AI chatbot for ecommerce?
An AI chatbot for ecommerce is software that answers shoppers' questions and handles tasks through chat, using a large language model trained on your store's products, policies and order data. It typically sits on your website, and often on email, SMS and social messaging too.
Not every chatbot works the same way. There are three broad types:
| Type | How it works | Good for |
|---|---|---|
| Rule-based bot | Follows scripted decision trees and menus | Simple, predictable questions, such as store hours |
| AI answer bot | Uses AI to understand questions and answer from your help center, catalog and policies | Product questions, sizing, shipping and returns policies |
| AI agent | Answers questions and takes actions in your systems, such as looking up orders or issuing refunds | Order tracking, returns, cancellations, address changes and subscription changes |
Most of the leading platforms for ecommerce brands now offer the third type, usually called AI agents, which can resolve a request completely rather than just pointing the customer to a help article.
What AI chatbots do for online stores
AI chatbots handle three main jobs.
Helping shoppers buy. Before a purchase, a chatbot can answer product questions, compare options, recommend sizes and suggest alternatives when something's out of stock. Shopify's free Inbox app now includes an AI sales associate that learns from your catalog, policies and store content to do exactly this.
Supporting customers after they buy. Much of an online store's support volume comes after the purchase: order tracking, delivery problems, returns, exchanges and address changes. These are repetitive, data-driven questions that AI agents can answer by checking the order itself.
Bringing customers back. Some tools, such as Klaviyo's Customer Agent, which became generally available in September 2025, combine support with marketing data. They can answer questions about delivery, sizing, order status and returns, and let shoppers add items to their basket and complete purchases.
The largest retailers show how big shopping assistants can get. Amazon says its Rufus assistant was used by more than 300 million customers in 2025 and helped deliver nearly $12 billion in incremental annualized sales.
What customers think of AI chatbots
Customers use AI for help more than ever, but not always the chatbot on your website.
3x
How much more likely customers are to use ChatGPT, Gemini or Copilot than a company's own chatbot for service issues
87%
Customers who say companies using AI for service must offer a way to reach a human
50%
Customers who say their interactions are easier when companies use generative AI
In Gartner's 2026 survey of 3,566 customers, use of company-provided chatbots was statistically unchanged since 2022. Customers are turning to AI for help, but often not to the chatbot on a brand's own website.
They have reason to be cautious. Qualtrics found that nearly one in five consumers who used AI for customer service saw no benefit, a failure rate almost four times higher than for AI use in general.
Usage is growing anyway, especially at busy times. Over the 2025 holidays, Salesforce found shoppers used retailers' own AI and agents for customer service 126% more than in the two months before.
How AI chatbots affect sales
Vendors and retailers often report that shoppers who use AI chat buy more. Amazon says US customers who use its Alexa for Shopping assistant, which combines Rufus and Alexa+, spend over 40% more per order than those who don't.
Treat figures like this carefully. Shoppers who start a chat are usually more interested in buying to begin with, so comparing chat users with everyone else overstates what the chatbot itself achieved. The only reliable way to measure the effect on your store is to compare similar visitors with and without access to the chatbot.
Popular AI chatbot platforms for ecommerce
These are some of the AI chatbot and agent platforms most used by ecommerce brands, with pricing as listed in October 2026:
| Platform | Best for | How it's priced |
|---|---|---|
| Shopify Inbox | Shopify stores that want a free AI sales associate for pre-purchase questions | Free |
| Gorgias AI Agent | Shopify brands that want a helpdesk and AI agent in one tool | Helpdesk plan plus $0.85 to $1.00 per automated interaction, depending on plan |
| Intercom Fin | Brands that want outcome-based pricing, on Intercom or another helpdesk | $0.99 per outcome |
| Zendesk AI agents | Larger teams already on Zendesk | Per-agent plans include 5 to 10 automated resolutions per agent a month, then $1.50 each (committed) or $2.00 (pay as you go) |
| Tidio | Small stores that want a low-cost starting point | Plans from $24.17 a month billed annually, including 50 one-time conversations with its Lyro AI agent |
| Siena AI | Mid-sized DTC brands that want an AI agent on top of their existing helpdesk | $750 a month platform fee plus $0.90 per automated ticket |
| Klaviyo Customer Agent | Brands on Klaviyo that want support and marketing data in one place | Priced by AI conversations |
Prices change often, so check the vendor's pricing page before you commit.
How AI chatbot pricing works
Most AI chatbot platforms now charge for results rather than seats. That sounds simple, but what counts as a result varies:
- Intercom Fin counts an "assumed resolution" when a customer doesn't reply for 24 hours after Fin's last answer.
- Gorgias counts an interaction as automated if the customer doesn't need a human agent within 72 hours.
- Zendesk changed its reporting in May 2026. Its automated resolution rate now includes "contained" conversations as well as "verified" resolutions, though it bills only for verified ones.
Two consequences follow. First, vendors' headline resolution rates aren't comparable with each other. Second, a customer who gives up and leaves can count as a success. Before you sign up, check how the platform defines a resolution and whether it matches what you'd count.
A quick way to estimate cost: if your store handles 2,000 conversations a month and an AI agent fully resolves half of them, at $0.99 per resolution you'd pay about $990 a month, plus any helpdesk subscription. Compare that with what those 1,000 conversations currently cost in agent time.
The risks of AI chatbots
Wrong answers. AI chatbots can confidently state things that aren't true. In 2024, a Canadian tribunal ordered Air Canada to compensate a customer after its chatbot gave incorrect information about a fare policy. The airline argued the chatbot was responsible for its own actions, which the tribunal called "a remarkable submission". Your chatbot speaks for your business.
Trapping customers. If customers can't reach a person, a chatbot becomes a wall. That's the main reason most customers want a human option.
Cutting too deep. Klarna's AI assistant handled two-thirds of its customer service chats in its first month in 2024, doing the work of 700 full-time agents. The following year, its CEO said the company's AI-driven cost-cutting in customer service had "gone too far" and planned a hiring drive so customers would always have the option of speaking to a real person.
Disclosure. In the EU, the AI Act's transparency rules, which apply from August 2026, require AI systems that interact with people to make clear they're AI, unless it's obvious. Even where it isn't required, being upfront avoids surprising customers.
How to set up an AI chatbot that customers like
Start with your most common questions. Look at your last few months of tickets. Order tracking, returns and sizing questions often make up a large share, and they're among the easiest to automate well.
Ground every answer in your own data. Connect the chatbot to your catalog, order system, help center and policies, and keep them up to date. Most wrong answers come from missing or outdated information.
Limit what it can do at first. Let the AI look up orders and answer questions before you let it issue refunds or cancel orders. When you do add actions, set limits and have the customer confirm before anything changes.
Always offer a way to reach a person. Make handoff easy and pass the conversation history along, so customers don't have to repeat themselves.
Test on real questions. Before going live, run the chatbot against a sample of past tickets and check the answers, especially for policies and edge cases.
How to measure an AI chatbot's performance
Track a few numbers from day one:
- Resolution rate, using your own definition rather than the vendor's.
- Re-contact rate: how often customers come back about the same issue within a few days.
- Customer satisfaction on AI conversations specifically, compared with human ones.
- Cost per resolution compared with the cost of a human-handled ticket.
- Conversion and order value, measured against a comparable group of visitors who didn't use the chatbot.
If resolution rates look high but re-contact rates and satisfaction scores are poor, the chatbot is ending conversations rather than solving problems.
Final thoughts
AI chatbots have become genuinely useful for online stores, particularly for repetitive post-purchase questions and pre-purchase product help. The tools are affordable, and many integrate with Shopify in minutes.
The brands that get the most from them treat the chatbot as part of their customer experience, not a way to avoid customers. Keep answers accurate, make a human easy to reach and measure the results honestly. For the bigger picture of how AI is changing shopping, see our guide to agentic commerce and our AI in ecommerce statistics.
