An AI chatbot for ecommerce is a software tool that answers customer questions and helps online stores manage routine support tasks. It can help shoppers find products, track orders, and get information about shipping and returns without waiting for a support agent.
What a store chatbot may do alone, with approval, and never
Hover or tap a tier or a panel below it for details.
For stores that receive the same customer questions every day, a chatbot can reduce repetitive work and help employees focus on requests that need personal attention. To work effectively, it needs accurate product information, access to relevant store systems, and clear rules about which tasks it can handle independently.
This guide explains what an ecommerce chatbot can do, the features it needs, and which tasks require human approval. It also covers when to choose a ready-made chatbot, when custom development makes sense, and how much each option costs.
1. When Your Store's Chatbot Speaks for Your Business
When customers ask your chatbot about a product, delivery, or return, they expect the information to be correct. They may use its answers to decide whether to place an order or request a service. Since the chatbot represents your store, its responses should match the information your support team provides.
For example, a customer asks whether they can exchange a product after 30 days. The chatbot should check the store's exchange policy and explain the conditions. If the request needs special approval, it should transfer the conversation to a support employee instead of promising an exchange.
The same applies to product prices, stock availability, discounts, and delivery dates. The chatbot should use information from your store's systems and policies rather than give answers it cannot confirm.
Store owners also need to keep this information updated. When products, prices, or policies change, the chatbot should use the latest details. Your team should regularly review its conversations to identify incorrect answers and requests that still require employee assistance.
The goal is to help customers get accurate information while reducing repeated questions for your support team. Track how many requests the chatbot resolves correctly and how often customers still need employee help. This will show whether it is improving your customer support process.
2. What an AI Chatbot Can Do for an Online Store
An ecommerce chatbot can help customers choose products, get answers during checkout, and receive support after placing an order. Businesses can connect it to their product catalog, inventory, order management system, and customer support software to handle different requests.
What it can do depends on the information and system access available to it. A chatbot connected only to your product catalog can answer product questions. One connected to your order management system may also provide delivery updates or help with returns.
Here are six ways online stores can use AI chatbots:
- Answer customer questions: Customers can ask about product sizes, availability, shipping charges, delivery times, and return policies. The chatbot answers using the store's information.
- Help customers find products: It can ask about a customer's needs, budget, or preferences and suggest matching products. For example, Shopify Inbox has an AI agent that uses product catalogs and store information to answer questions and recommend products.
- Track orders and deliveries: Customers can check their order status and delivery updates through the chat. The chatbot should verify their identity before sharing private order details.
- Assist with returns and exchanges: It can check whether an order meets the return policy, collect the required information, and guide customers through the process. Requests that need approval can be passed to an employee.
- Support customers during checkout: The chatbot can explain payment options, shipping charges, and available discounts. Some tools also support abandoned-cart follow-ups through approved marketing channels.
- Transfer complex requests to support employees: When a customer needs additional help, the chatbot can transfer the conversation and relevant order information to a support agent.
These features help online stores answer customer questions faster, reduce repetitive support work, and assist shoppers throughout the buying process. For stores receiving large numbers of similar requests, a chatbot can also help the existing support team manage more conversations.
For example, Orthofeet, a US footwear brand, uses Gorgias AI Agent for order tracking and returns. According to Gorgias's Orthofeet case study, the company automated 56% of its support tickets in less than two months.
Cocorico, a French clothing brand, uses the same tool for delivery updates, sizing questions, returns, and exchanges. The company operates with a two-person support team. Gorgias reports that Cocorico automated 48% of its support tickets in about three months.
Gorgias reports these results, and actual performance will depend on the store's support volume, customer requests, and available integrations.
For ecommerce businesses, the main benefit is handling more customer requests without increasing the support team's manual workload. Employees can spend more time resolving complex issues, while customers receive help with common questions throughout their shopping experience.
3. When an Online Store Doesn't Need an AI Chatbot
Not every online store needs an AI chatbot. If your business receives only a few support messages each day and employees already answer them on time, a shared inbox, saved replies, and clear help pages may be enough.
The type of customer request also matters. Some stores receive mostly complaints about missing parcels, damaged products, or payment disputes. These problems often require someone to investigate and make a decision. A chatbot may collect the details, but employees would still handle most of the work.
Before buying a chatbot, review one month of customer messages. Check which questions repeat, how much time employees spend answering them, and how many requests could be completed without staff involvement.
You should also check whether your store information is ready. A chatbot cannot give reliable stock updates if inventory records are incorrect. It also needs access to current order information before it can answer delivery questions.
Start by improving the support process you already have. Update product pages, make return policies easy to find, and use automatic order notifications where possible.
Then compare the cost of a chatbot with the employee time it could save. If support volume is low, the subscription and setup costs may be higher than the value of the work it removes.
You can review the decision again when customer messages increase, response times become longer, or employees spend more time answering the same questions.
4. What a Good Ecommerce Chatbot Needs
Customers may ask about a product before buying and contact the store again to check their order. A good ecommerce chatbot should handle these requests using accurate information and the right system connections.
Before buying a tool or developing a custom chatbot, check whether it has the following features.
Up-to-Date Product and Store Information
The chatbot needs accurate product descriptions, prices, stock levels, shipping details, and return policies to answer customer questions.
Store employees should be able to update this information when products, prices, or policies change. For example, if a store introduces a longer return period for holiday orders, the chatbot should explain the new policy instead of using the old one.
Connection to the Order Management System
Customers often want to know whether their order has been confirmed, shipped, or delivered. Connecting the chatbot to your order management system allows it to retrieve this information.
The chatbot should check the latest order status rather than estimate it. It should also verify the customer's identity before showing private order details.
Conversation History and Context
Customers often ask more than one question during a conversation. The chatbot should understand previous messages so customers do not need to repeat themselves.
For example, a customer may ask whether a pair of shoes is available in size 8 and then ask if it comes in black. The chatbot should understand that both questions refer to the same product.
Transfer Conversations to Support Employees
Some requests need help from a support employee. A customer may report a missing parcel, question a payment, or request an exception to the return policy.
The chatbot should transfer these conversations along with the customer's questions, relevant order details, and previous messages. This helps employees continue the conversation without asking customers to repeat everything.
Stores should also decide what happens when customers request human support outside regular working hours.
Controlled Access to Customer Information
A chatbot should only have access to the information it needs for its assigned tasks.
For example, product recommendations require access to the product catalog. Order tracking requires access to the relevant order record after the customer has been verified.
Developers should set these permissions in the connected software. Store owners should also check where customer conversations are stored, who can access them, and how long the information is kept.
Support Across Customer Communication Channels
Customers contact online stores through websites, WhatsApp, Instagram, and other platforms.
A chatbot that supports these channels can help businesses manage questions from different places. Before choosing a tool, check which channels it supports and whether employees can manage conversations through one support inbox.
Some providers charge extra for additional communication channels, so this should also be included in the cost comparison.
Testing and Regular Improvements
Before launching, test the chatbot with questions from previous customer conversations. Include product availability, delivery updates, return requests, and situations where order information is missing.
Check whether the chatbot gives the correct answer, asks for the right details, or transfers the conversation when it cannot help.
After launch, review conversations regularly. Wrong answers, repeated customer questions, and failed requests can show where the chatbot needs better information or changes to its setup.
Performance Tracking and Reporting
Store owners need to know whether the chatbot is actually reducing support work.
Useful measures include the number of requests resolved without employee help, the number transferred to support agents, response times, and the time employees save on routine questions.
Intercom reported an average 76% resolution rate for Fin across its customers. Tidio's Jiffy.com case study reports a 70% resolution rate using Lyro.
Both figures come from the chatbot providers, and their methods of counting resolved conversations may differ. These results are useful examples, but stores should measure their own performance.
Vendor-reported results, grouped by what they measure
Hover or tap a result to see who reported it and what was counted.
A resolved conversation should mean that the customer received the help they needed. If the chatbot answers a question but the customer contacts an employee again to finish the task, the store has not necessarily saved the full amount of work.
5. What a Chatbot Can Do Alone, What Needs Approval, and What It Must Never Do
A chatbot that only answers questions has limited access to your business. Once it connects to order, customer, or payment systems, it may also be able to take actions.
Some actions are safe to automate. Others need an employee to check and approve them. Store owners should decide these limits before allowing the chatbot to make changes in connected systems.
What the Chatbot Can Handle Alone
A chatbot can answer product questions, check stock levels, explain shipping policies, and provide order updates using information from your store.
For example, a customer asks whether their order has shipped. The chatbot verifies the customer's identity, checks the order management system, and shares the latest shipping status.
It can also collect return information, create support tickets, and guide customers through the return process.
These tasks can be automated when the chatbot has the required information and permissions. Developers should decide which systems it can read and which actions it can complete without employee involvement.
What Needs Human Approval
Some customer requests involve refunds, payments, discounts, or changes to an order. These actions can affect the store's money or customer records.
For example, a customer requests a refund. The chatbot can check the order, review the return policy, and collect the required details.
If the store allows automatic refunds within certain limits, the connected software can check whether the request qualifies. A refund outside those limits should go to a support employee before any payment is made.
The same applies to unusual discounts, payment disputes, and changes to sensitive account information. The chatbot can prepare the request, but an authorized employee should decide whether to approve it.
This allows stores to automate routine work while keeping important decisions under employee control.
What the Chatbot Must Never Do
A store chatbot should not be able to access or change information outside its assigned tasks.
It must not show one customer's private order details to another customer. It should not reveal confidential business information or change product prices, payment terms, or store policies without authorization.
It also should not promise discounts, refunds, or transactions that the store has not approved.
A real incident shows why these limits matter. In December 2023, someone asked a chatbot on a Chevrolet dealership's website to agree to sell a vehicle for $1. The chatbot agreed. No sale took place, but the exchange was recorded as AI Incident Database incident 622.
The problem was that the chatbot could be persuaded to make a promise outside the dealership's actual offers.
For an online store, prices, refunds, and order changes should be controlled by the store's software. The chatbot should not be able to bypass those rules, regardless of what a customer asks it to do.
How to Control What Your Chatbot Can Do
These limits need to be built into the software connected to the chatbot. Instructions telling it not to perform certain actions are not enough.
For example, suppose a customer asks to change the delivery address of an order.
- Customer request: The chatbot collects the new address and the order number.
- Identity check: The system confirms that the customer is authorized to make changes to that order.
- Order check: The store's software checks whether the order has already shipped and whether an address change is still allowed.
- Action or approval: If the change is permitted, the system updates the address. Otherwise, it sends the request to an employee.
- Confirmation and records: The system checks whether the update succeeded, tells the customer what happened, and keeps a record of the action.
The chatbot communicates with the customer, but the connected software decides whether the requested change is allowed.
Changing a delivery address: the chatbot talks, the software decides
Hover or tap a step to see what happens there.
CodingCrafts used similar controls while developing an AI-powered financing platform. The assistant could negotiate financing amounts, rates, and terms with merchants, but developers set limits on the offers it could make.
The assistant also followed the stage of each funding application. It could not negotiate offers for declined applications or discuss offers before the application reached the correct stage.
An ecommerce chatbot can use the same approach for refunds, order changes, and other sensitive tasks. The business sets the rules, and the software checks them before allowing an action.
6. Buying an Ecommerce Chatbot: Comparing Ready-Made Tools
Online stores can use ready-made chatbots for product questions, order updates, and customer support. Some tools also help shoppers find products and complete purchases.
The right option depends on your ecommerce platform, monthly support volume, and the tasks you want to automate.
The following comparison uses published features and pricing from five providers. It is not based on hands-on testing.
| Tool | Best suited for | Published pricing | What to check |
|---|---|---|---|
| Shopify Inbox | Shopify stores needing customer chat and product recommendations | Free Shopify app with an AI agent that uses store information | Whether it supports the order-related tasks your store needs |
| Tidio Lyro | Small and growing Shopify or WooCommerce stores | $32.50/month for 50 AI conversations to $583.33/month for 1,000, billed annually | Order integrations and conversation limits |
| Gorgias AI Agent | Ecommerce stores with frequent order and return requests | Helpdesk ticket fees plus AI automation charges under applicable plans | Total ticket and automation charges |
| Intercom Fin | Businesses needing shopping assistance, customer support, and connected workflows | From $0.99 per qualifying outcome, with plan charges where applicable. Advanced is $85 per seat/month on annual billing | Which outcomes are billable and what integrations are available |
| Zendesk AI | Larger support teams managing several communication channels | Suite Team starts at $55 and Suite Professional at $115 per agent/month, billed annually | Included automated resolutions, usage charges, and integrations |
Prices are in USD and were checked against published vendor information on November 18, 2026. Features, plans, and billing rules can change.
Sources: Shopify Inbox, Tidio, Gorgias, Intercom, Zendesk.
How Chatbot Pricing Affects Your Monthly Cost
Chatbot providers do not all charge for the same activity.
Tidio counts a Lyro conversation when the AI replies, even if it does not fully resolve the issue. Intercom charges for qualifying outcomes, which can include certain completed handoffs. Gorgias may charge both a helpdesk ticket fee and an AI automation fee when its agent fully resolves a conversation, depending on the billing plan. Zendesk uses automated-resolution pricing alongside its support plans.
This makes monthly support volume important when comparing tools.
Suppose your store receives 1,000 customer messages each month. Some may be simple product questions, while others involve refunds, missing orders, or payment problems. The chatbot may resolve some requests and transfer others to employees.
Your bill will depend on which requests the chatbot handles, how the provider counts them, and which features you use.
Ask each vendor to estimate the total monthly cost using your store's actual support activity. Include subscription charges, usage fees, additional communication channels, and any setup or integration work.
What to Check Before Buying an Ecommerce Chatbot
1. Review Your Customer Support Requests
Look at one month of customer messages. Identify common questions about products, deliveries, returns, and payments.
Count how often they appear and how long employees take to handle them. This helps you identify which tasks are worth automating first.
2. Check the Available Integrations
Find out whether the chatbot connects to your product catalog, inventory, order management system, and support software.
For example, if customers frequently ask about deliveries, the chatbot needs current order information. Ask the vendor to demonstrate how it checks an order and shares the result.
3. Test the Tools with Real Customer Questions
Select 20 to 30 questions from your support inbox. Remove names, addresses, payment details, and other private customer information before testing.
Use the same questions with each shortlisted tool. Include product inquiries, order tracking, returns, and requests that need employee approval.
Check whether answers are correct, how quickly the tool responds, and whether it transfers difficult requests with the necessary details.
4. Compare the Total Cost
Check the monthly subscription, included conversations, extra usage charges, and setup requirements.
A tool may fit your subscription budget but require additional development to connect to your systems. Include that work when comparing options.
5. Start with One Support Channel
Launch the chatbot where most customers contact your store, such as website chat.
Review conversations during the first few weeks. Check how many requests it resolves correctly, how much employee time it saves, and what the service costs.
Once it works reliably, you can add more tasks and channels.
When to Consider a Custom Chatbot
Ready-made chatbots work for many common ecommerce tasks. However, some stores use several systems or have business rules that standard tools cannot fully support.
For example, an online store may use Shopify for orders, a separate warehouse system for shipping, and another application for returns. An address change could require checking and updating more than one system.
If the chatbot can only collect the customer's request, an employee may still need to complete the rest of the work.
Before investing in custom development, check whether a ready-made tool can support the process through its existing features or integrations. If it cannot, a custom chatbot may be worth considering.
7. When a Custom Ecommerce Chatbot Is Worth the Cost
A custom chatbot becomes useful when standard tools cannot complete an important support task or follow the rules your business needs.
Consider a store that uses Shopify for orders, two warehouses for inventory, and a separate application for returns.
A customer wants to exchange a product for another size. To complete the request, the chatbot needs to check the original order, confirm the return period, find the replacement item in stock, and determine which warehouse can supply it.
If the exchange is allowed, the system may need to create a return request, reserve the replacement product, and update the connected applications.
One size exchange, three systems, two ways to handle it
Switch views, then hover or tap a step or a system for details.
A chatbot that only answers questions cannot complete this process. Employees would still need to open several systems and make the changes manually.
Custom development can connect these systems and automate the steps that follow clear business rules. The same need can arise with wholesale prices, subscription orders, split shipments, or return policies that differ between countries.
How CodingCrafts Has Solved a Similar Integration Problem
CodingCrafts worked on a similar business problem while developing an AI-powered CRM system.
The client managed sales leads through calls, WhatsApp, SMS, and email. Employees had to check different tools, update customer records, and follow up on inquiries manually. This made it difficult to respond quickly and keep track of conversations.
Our developers used Twilio to connect these communication channels and built one system for managing leads and customer interactions. We also added automatic lead assignment, follow-up workflows, and AI voice agents for first contact.
According to the published CodingCrafts CRM case study, first response time fell from 2–6 hours to under one minute. Manual workload decreased by approximately 50–60%.
These results come from a sales CRM project, not an ecommerce chatbot. They show how connecting business systems and automating repeated tasks can reduce work that employees previously handled manually.
A custom ecommerce chatbot can follow the same development approach. Businesses do not need to train an AI model from scratch. Developers can use an existing model and build the necessary software around it.
The work may include connecting order and inventory systems, checking customer identity, setting permissions, applying return policies, and recording important actions. The team must also test how the chatbot responds when information is missing or a connected system fails.
Custom development is not always the better choice. If a ready-made chatbot can complete the required tasks at a lower total cost, it may be enough.
Before paying for a complete custom solution, choose one support task that takes employees considerable time. A small proof of concept can test whether the chatbot completes it correctly, follows the store's rules, and reduces manual work.
The results can help you decide whether further development is worth the investment.
8. What an Ecommerce Chatbot Costs: Ready-Made Tools vs. Custom Development
Ready-made chatbots usually cost less to set up than custom solutions. Businesses pay a subscription, and some providers charge extra for conversations, automated tasks, or support seats.
As shown in Section 6, providers use different pricing models. Check what your store would actually pay at its monthly support volume rather than comparing only the advertised starting prices.
Custom development requires a larger upfront budget because developers build the chatbot around your store's processes. This may include connecting order systems, warehouse software, customer records, and payment services. Developers also create the rules that control what actions the chatbot can take.
According to CodingCrafts' published AI development estimates, a focused proof of concept usually costs $10,000–$30,000. Production AI systems typically range from $50,000 to $250,000 or more.
These are general AI development planning ranges, not fixed prices for ecommerce chatbots. The actual cost depends on the number of systems involved, the work required to prepare business data, and how much testing and monitoring the solution needs.
What Adds to Custom Chatbot Costs?
The development budget is only one part of the total cost.
Businesses also need to consider ongoing expenses such as AI model usage, hosting, third-party services, monitoring, and maintenance. A chatbot that handles more conversations or makes frequent calls to connected systems may have higher running costs.
Support requirements also matter. If the store changes its return process, replaces its order software, or adds a new delivery partner, the chatbot may need updates.
Ask developers to separate the initial build cost from expected monthly operating costs so you can plan for both.
Two cost shapes: a ready-made tool and a custom build
Hover or tap a cost to see what it covers and the published prices.
How to Decide Which Option Is Worth Paying For
The better investment depends on how much work the chatbot can complete. If customers mostly ask about product availability, shipping charges, and order status, a ready-made tool may be enough. It can handle common questions without a large development budget.
Custom development becomes worth considering when employees still have to complete several steps after the chatbot responds.
For example, a store may use separate systems for orders, warehouse operations, and returns. If an employee must open all three applications to process an exchange, answering the customer's question removes only a small part of the work.
A custom solution could connect those systems so the chatbot checks the order, follows the return policy, and prepares or completes the exchange within approved limits.
Before investing, review your support records and calculate how much employee time each type of request takes. Focus on tasks the chatbot could complete correctly from start to finish, not just the number of messages it could answer.
You can estimate the time saved by multiplying the number of requests it can fully handle by the average employee time required for each request. Then compare the expected savings with subscription fees or development and operating costs.
If the process involves several systems or complex rules, test one workflow first. This gives you a clearer picture of what the solution can do and what it may cost before committing to a complete build.
9. Build an Ecommerce Chatbot That Works with Your Store's Systems
A ready-made chatbot may be enough for product questions, order tracking, and basic support. Stores that use several business systems or spend significant employee time on returns, order changes, and other repeated tasks may benefit from custom development.
CodingCrafts develops AI agents and business software that connect with existing systems and follow defined business rules. Our work includes understanding workflows, designing integrations, setting permissions, testing automated actions, and monitoring systems after launch.
Book a free 30-minute discovery call with CodingCrafts to discuss your current support process, ecommerce integrations, and development budget. We'll help you assess whether a ready-made tool can meet your needs or a custom chatbot is worth building.
10. Frequently Asked Questions
How Long Does It Take to Set Up an Ecommerce Chatbot?
A ready-made chatbot can often be set up in a few hours if your product information and store policies are ready. More time may be needed to test responses, configure integrations, and prepare employee handoffs.
A custom chatbot may take several weeks or longer, depending on the features and business systems involved. Businesses can start with common customer questions and add order tracking, returns, and other functions after testing.
Do Customers Have to Be Told They Are Talking to a Chatbot?
Some laws require businesses or AI providers to tell people when they are interacting with a bot.
In California, Business and Professions Code § 17941 prohibits using a bot to intentionally mislead someone about its identity in order to deceive them in certain commercial transactions or elections. Clear disclosure that the system is a bot addresses the disclosure requirement under that section.
In the European Union, Article 50 of the AI Act includes transparency requirements for AI systems that interact directly with people. These rules began applying on August 2, 2026. Providers must design such systems so people are informed they are interacting with AI, unless this is already obvious.
The exact obligations depend on the system and how it is used. Businesses should review the rules that apply to their operations. A practical approach is to identify the chatbot clearly and make it easy for customers to contact a person.
Will an AI Chatbot Replace My Customer Support Team?
An AI chatbot can reduce the time employees spend answering repeated questions, but it does not remove the need for human support.
In the examples covered in Section 2, Orthofeet and Cocorico used automation to handle more customer requests without expanding their support teams at the same rate.
Employees are still needed for complicated complaints, payment disputes, unusual return requests, and situations that require approval.
The main benefit is allowing employees to spend less time on routine questions and more time helping customers with problems the chatbot cannot resolve.
Does an AI Chatbot Work with Shopify?
Yes. Shopify stores can use Shopify Inbox, which includes an AI agent that answers questions using product details, store policies, and other approved information.
Other tools, including Tidio Lyro and Gorgias AI Agent, also support Shopify stores.
However, answering an order question and changing an order are different tasks. Before choosing a tool, check whether it can perform the exact Shopify actions your store needs, such as checking delivery status, starting a return, or updating an order.
What Languages Can an Ecommerce Chatbot Handle?
It depends on the tool and the languages it supports.
Some chatbots can answer customers in several languages, but they still need correct product information, shipping terms, and return policies.
If your store serves customers in different countries, test the chatbot in the languages they use. Check whether it explains product details and policies correctly rather than relying only on automatic translation.
What Is the Difference Between an AI Chatbot and an AI Agent?
An AI chatbot mainly answers questions and guides customers using available information.
An AI agent can also take actions in connected software when it has permission. For example, it may check an order, confirm return eligibility, and submit a return request.
Some ecommerce tools combine both capabilities.
The main difference is what the system is allowed to do. The more actions it can perform, the more important it becomes to control access, check business rules, and require approval where needed.
What Data Does an Ecommerce Chatbot Need, and Is It Safe to Give It Access?
A chatbot that answers common questions needs product details, prices, shipping information, return policies, and other help content.
For order tracking, it may also need access to order records after verifying the customer's identity. It should not have access to information or actions unrelated to its tasks.
Before connecting a chatbot, ask the provider where conversations are stored, who can access them, how long the data is kept, and whether customer information is used to train AI models.
Customer information and sensitive actions should be protected through permissions in your store's software. The chatbot should only have the access required to complete its assigned work.
Not Sure Whether to Buy a Chatbot or Build One?
Tell us about your store's systems, support volume, and return rules. CodingCrafts will help you decide whether a ready-made tool covers them or a custom chatbot is worth building.
