Back to blog

· askVela-Team

AI Chatbot Use Cases in Ecommerce: A Shopify and WooCommerce Playbook

An order-status question and a product-compatibility question need different data and safeguards. Use this playbook to decide what your store's chatbot can safely answer.

Updated:

The useful AI chatbot use cases in ecommerce are the ones your store can support with a reliable source. A shipping-policy question needs approved text; product advice needs verified catalog details; an order-status answer needs access to the right order. For a Shopify or WooCommerce store selling in Europe, decide what the bot may say before you enable each task.

Match each chatbot use case to the information it needs

Use this matrix as a launch gate. If the required source is missing or outdated, provide a route to a person instead of trying to improve the prompt.

Stage and questionRequired sourceSafe responseHandoff trigger
Pre-purchase: “Where do you deliver?”Approved shipping policyGive published destinations and estimates, without promising a date.Destination or requested delivery promise is not documented.
Pre-purchase: “Can I return this?”Current returns policy and applicable product informationExplain the published process; do not decide a disputed case.Formal withdrawal request or disputed eligibility.
Product discovery: “Will this fit?”Current catalog specification or verified compatibility guideSuggest a product only when the source supports the match.Compatibility is missing, ambiguous or contradictory.
Checkout: “Why was my code rejected?”Published promotion termsAsk what message appeared and explain any relevant published condition.The terms do not explain the error, or technical diagnosis is needed.
Post-purchase: “Where is my order?”Connected shop order data and a permitted lookupAfter checking the required lookup details, report only available order and shipping fields.Lookup fails, details do not match or delivery is disputed.
Post-purchase: “Change my address” or “Refund me”Verified action workflow and authority to actAcknowledge the request and collect only context needed for review.No verified workflow supports the action, or the case needs a decision by a person.

Answer pre-purchase questions without guessing

Map each answer to an approved source: shipping destinations and estimates to your shipping policy, return instructions to your returns policy, sizing to the relevant size guide, and payment methods to your current checkout information. Claims about a particular product’s material or features need product-specific evidence, not a general policy.

Give each source a content owner. When a policy changes, have that owner update the source and rerun the questions it answers; an old answer does not become safe just because the bot repeats it consistently. If your sources need organizing, build and maintain your store’s knowledge base around actual customer questions.

Guide product discovery using verified catalog details

Ask about intended use, model or fit, and essential features, then compare the answers with the catalog. In WooCommerce, check that the product fields used for advice contain the specifications customers actually need. A shared category or similar product name is not evidence of compatibility.

Keep fit and availability separate. “Available to order” does not establish compatibility. State an exact stock quantity only if current shop data supplies it and you have enabled that disclosure; do not assume a quantity appears on a product card. If you use askVela’s product recommendations with its WooCommerce integration, check the underlying product data before allowing technical recommendations.

Resolve checkout questions, not unverified checkout errors

Published promotion terms, accepted payment methods and shipping options can support checkout answers. An unexplained error cannot: without evidence from the checkout, the bot should not claim to know why a code or payment failed.

Test this with a code that fails a documented condition and with an error the terms do not explain. The first can receive a source-backed explanation; the second needs a handoff. Use the same boundary for payment failures.

Handle order status with the right Shopify or WooCommerce connection

“Where can I find my tracking link?” can be answered from approved help content. “What is the status of my order?” requires a permitted lookup against shop data. askVela supports order lookup through its Shopify integration and WooCommerce plugin; its lookup can be configured to use email alone or order number plus email. Choose and test the lookup mode and the fields you intend to disclose.

  1. Check which identifiers the configured lookup requires. For an order-number-and-email lookup, require both before attempting it.
  2. Return customer-specific details only after the lookup matches. If it fails, reveal no status, carrier or order contents.
  3. Report only fields the shop returns, such as order status, shipping status, carrier or tracking link when available. A tracking link is not a parcel-scan history.
  4. Route address changes and disputed deliveries to a person; do not promise that the bot has edited an order.

This example assumes that the lookup matches and those fields are returned. If either identifier is wrong, the response should say that the order could not be verified, without confirming whether the other identifier belongs to a customer.

Set human handoff rules before launch

Write rules around observable customer statements and failed checks. An agent needs the customer’s request, the relevant product or order reference, the source consulted and any approved steps already tried—not card details or unrelated personal information.

For a formal withdrawal, acknowledge the request and route it for handling; do not substitute a generic returns-policy answer for the customer’s stated intent. Define troubleshooting steps in advance rather than letting the bot improvise a payment diagnosis. With askVela, you can preview custom guidance using bot rules and test console. Test both a customer-requested handoff and a suggested handoff, including what the customer sees before a person takes over.

Test the use cases before enabling them for EU customers

EU hosting does not complete your store’s privacy assessment. Check what the widget collects, what an order lookup may disclose, who can access the conversation and what your store tells customers. askVela has encrypted transmission and privacy controls; its security and GDPR controls are inputs to your review, not a substitute for it. Separately test your chat widget’s consent setup.

Enable a use case only after its source, safe answer and handoff tests pass. If a source changes, rerun the relevant tests before relying on the answer again.