Skip to content
CentriWeb
Not sure where to start?

Tell us how the business runs. We'll tell you what we'd build first.

Book a call
For builders and tradesPreBuildMore homeowner enquiries, more trust, more jobs won. For sole tradersOwner CFOA financial operating system for owners who do the books.
Need something of your own?

We build custom software too.

Talk to us

Chatbot for ecommerce: order questions, returns and sales

What an ecommerce chatbot can answer, how order lookup connects to Shopify and WooCommerce, when it hands over to a person and when to skip it.

A chatbot for ecommerce answers the questions that fill a store’s inbox: where is my order, how do I return this, does it come in another size. It can look up an order, explain your returns policy and answer product questions from your catalogue, and it passes anything it cannot settle to a person. The goal is fewer routine emails and more visitors getting an answer before they leave without buying.

What a chatbot can handle for an online store

Store owners usually find that a small number of questions make up most of the support load. A chatbot suits these because the answers already exist somewhere: in the order system, the returns policy or the product page.

  • Order status. The customer gives an order number and the email used at checkout, and the bot reads the status and tracking details from the store.
  • Returns and exchanges. It explains the policy in your words and starts the process you already run, whether that is a form, a returns label or an email to a person.
  • Product questions. Sizing, materials, compatibility and stock, answered from the catalogue and the product descriptions you wrote.
  • Delivery and shipping. Costs, countries you ship to and timeframes, as stated on your shipping page.

It should not guess. If the answer is not in your catalogue or policies, the right reply is “I will pass this to the team” and not a plausible-sounding invention.

Order lookup: where the data comes from

Order lookup is the feature that saves the most time, and it is also the one that needs the most care, because it connects the chatbot to customer data.

For a WooCommerce store, the platform’s own REST API lets an application read and manage orders, products and customers, including order notes and refunds. Access uses a consumer key and secret that you generate in WooCommerce under Settings, Advanced, REST API, and each key is tied to a WordPress user and set to read, write or both. Their documentation recommends HTTPS (WooCommerce REST API documentation, as at November 2026). A chatbot that only looks up orders needs a read-only key.

For a Shopify store, orders are available through Shopify’s Admin API. Shopify’s developer documentation says access depends on permission scopes, that access to protected customer data is required, and that by default an app can only read the last 60 days of orders unless extra access is granted. It also notes the older REST Admin API is a legacy API and that new public apps must use the GraphQL Admin API (Shopify Admin REST API, Order resource, as at November 2026). That is a technical detail for whoever builds the connection, but it explains why a ready-made chat app is often simpler than a custom build for a Shopify store.

Whichever platform you use, the chatbot should confirm identity before it shares anything about an order. Asking for the order number and the email address on the order is the usual minimum. It should never show a full address or payment details in the chat.

Returns, policies and product questions

The bot is only as good as the information you give it. Before launch, collect:

  • Your returns and refund policy, with the time limit, condition rules and who pays for shipping.
  • Your shipping rates, regions and cut-off times.
  • The product catalogue, including size charts, care instructions and anything customers often ask that is not on the product page.
  • A list of questions it must always hand to a person, such as damaged goods, payment disputes and anything involving a complaint or legal threat.

Keep this as one written document the whole team controls. When a policy changes, change the document and the bot’s answers follow. Chatbots that quote an old returns window cause disputes that cost more than the support time they saved.

For product questions, answers should come from your catalogue data and not from general knowledge. A bot that tells a customer a jacket is waterproof because most jackets of that type are, when yours is not, is a refund waiting to happen. Test it with the trickiest ten questions your support inbox has ever received.

How to add a chatbot for ecommerce to your store

There are three broad ways to add a chatbot to an online store, and the right one depends on your platform and how much custom behaviour you need.

  1. A chat app or plugin for your platform. Quickest to start, with a monthly fee that usually grows with usage. Check where customer data is sent and who can read the chat history.
  2. A chat platform you embed on the store. More control over the answers, with a connection you set up to the store’s order data.
  3. A custom build. The automation tool n8n has ready-made nodes for both stores: its Shopify node supports orders and products (create, get, update, delete) and its WooCommerce node covers orders, products and customers (n8n Shopify node, n8n WooCommerce node, as at November 2026). CentriWeb builds most client automations in n8n, so we have a preference here, and a plugin may be enough for a small store.

The hand-off to a person matters in all three. Make sure a customer who types “speak to someone” reaches a real inbox, phone number or live chat, and that the conversation travels with them.

Using the chatbot to complete more sales

A chatbot can help a hesitant visitor buy, but only in modest ways. The useful moments are:

  • A visitor on a product page asks about sizing or delivery time, and the bot answers in seconds instead of the visitor leaving to look it up.
  • A visitor in the cart asks about returns, and the bot confirms the policy.
  • A visitor asks for something you do not stock, and the bot suggests the closest alternative from the catalogue.

Avoid pop-ups that interrupt before the visitor has seen anything, and avoid pressure such as invented stock levels. Customers notice, and it damages trust in every other answer the bot gives.

How to tell whether it is working

A chatbot is a cost, so it should earn its place in the same way any other tool does: by saving time, bringing in more sales or keeping sales you would otherwise lose through a refund or an abandoned cart. If it does none of the three after a couple of months, change what it answers or switch it off.

Pick three numbers before launch and look at them monthly:

  • The number of “where is my order” and returns emails and calls, compared with the month before.
  • The share of chats resolved without a person, and the share handed over.
  • The conversion rate of visitors who chatted compared with those who did not. Treat this as a rough signal only, because people who chat may already be more interested.

Read a sample of real conversations every week for the first month. The mistakes are easy to spot and cheap to fix when you catch them early.

When a chatbot is not worth it

If your store gets a few support emails a week, a clear FAQ page and a fast reply may be a better use of your time. A chatbot also needs upkeep. Prices change, products are discontinued and policies shift, so someone has to own the information it answers from. If nobody will, leave it out.

If you want a chatbot connected to your store, your inbox and your CRM, built and maintained for you, CentriWeb does this as part of its AI answering and lead follow-up service. For a broader look at website chat, read our guides on an AI chatbot for a small business website and Shopify versus WooCommerce, and see our AI guides for more.

Common questions.

Can a chatbot look up customer orders?

Yes, if it is connected to your store's order data through the platform's API or an automation tool. It should ask for the order number and the email on the order before sharing anything, and should never show payment details.

Does a chatbot work with Shopify and WooCommerce?

Both platforms offer APIs that expose orders and products, and chat apps and automation tools such as n8n have connections for each. Check the current access rules in each platform's own documentation, as they change.

What should an ecommerce chatbot hand to a person?

Damaged or wrong items, refund disputes, payment problems, complaints and anything it cannot find in your policies or catalogue. A customer who asks for a person should reach one.

How do I stop a chatbot giving wrong answers about products?

Limit it to answers drawn from your catalogue, policies and shipping page, and tell it to hand over when the answer is not there. Test it with the hardest questions from your support inbox before launch.

Related guides.

AI for business · 5 min read

AI for plumbers: the jobs it does well today

Talk to CentriWeb about this

CentriWeb helps businesses put AI to work where it saves time or money, and keeps it away from the jobs it should not do. The first call takes 30 minutes and is free.

Or call +61 483 957 775 or email info@centriweb.com. CentriWeb is based in New Zealand and works online with businesses worldwide.

Ask a questionBook a call