Build an e-commerce support bot

Build an AI support bot for your online store that handles order status inquiries, processes return requests, recommends products, and recovers abandoned carts — turning your customer support into a revenue driver.

Overview


Goal	Deploy a Conversation AI bot that handles order status lookups, return initiation, product recommendations, and abandoned cart follow-up
Time to build	60–90 minutes
Prerequisites	Conversation AI enabled, knowledge base populated, workflows configured, e-commerce integration connected
AI features used	Conversation AI + Knowledge Base + Workflows
Difficulty	Intermediate

Create new bot dialog showing bot type options — The Create new bot dialog with General Q&A selected for e-commerce support

Architecture overview

The e-commerce support bot handles four primary flows:

Order status — Customer asks “Where is my order?” The bot looks up the order using the customer’s email or order number and provides tracking information.
Returns and exchanges — Customer wants to return or exchange an item. The bot verifies the return window, collects the reason, and initiates the process.
Product recommendations — Customer asks for help choosing a product. The bot asks about their needs and recommends items from your catalog.
Cart recovery — A workflow detects an abandoned cart and triggers the bot to re-engage the customer with a personalized message.

Step-by-step build

Connect your e-commerce integration

Navigate to Settings > Integrations and connect your e-commerce platform (Shopify, WooCommerce, or other supported platforms).This connection enables:

Order data syncing to contact records
Purchase history visibility
Abandoned cart detection
Product catalog access

If your e-commerce platform is not directly supported, you can use the HoopAI API or Zapier to sync order data to custom fields on the contact record.

Build the e-commerce knowledge base

Navigate to AI Agents > Knowledge Base and add:Store policies:

Return and exchange policy (timeframes, conditions, exceptions)
Shipping options and estimated delivery times
Warranty information
Price match or adjustment policy

Product information:

Product categories and descriptions
Sizing guides and fit recommendations
Material and care instructions
Frequently compared products

Order and shipping FAQs:

How to track an order
What to do if an order is damaged or missing
International shipping details
Gift wrapping and gift card options

Account and payment:

How to create or reset an account
Accepted payment methods
How to apply discount codes
Subscription management (if applicable)

Keep your knowledge base synced with your actual store policies. If you change your return window from 30 days to 14 days, the knowledge base must be updated immediately. Inconsistent information leads to customer complaints and potential chargebacks.

Create custom fields for e-commerce data

Navigate to Settings > Custom Fields and create:

Field name	Type	Purpose
Last Order Number	Text	Most recent order ID
Last Order Status	Dropdown	Options: Processing, Shipped, Delivered, Returned
Customer Lifetime Value	Number	Total spend to date
Return Count	Number	Number of returns initiated
Cart Abandoned	Checkbox	Whether an active cart was abandoned
Product Interests	Text	Categories or products the customer has browsed

Create the Conversation AI bot

Navigate to AI Agents > Conversation AI and create a bot:

Name: “E-Commerce Support”
Type: Auto-pilot
Channels: Web Chat (primary), SMS, Facebook Messenger, Instagram, WhatsApp
Calendar: Leave unassigned

See Conversation AI setup for detailed steps.

Write the bot prompt

Paste the full prompt from the section below. The e-commerce prompt handles multiple conversation types (order status, returns, product help, general questions) and routes between them based on the customer’s intent.

Configure bot actions

Set up Conversation AI actions:

Update custom fields: Record the order number, issue type, and resolution
Add tags: “order-inquiry,” “return-request,” “product-recommendation,” “cart-recovery”
Create task: For issues that need human intervention (e.g., damaged items, refund approvals)
Send internal notification: Alert the support team for high-value customers or complex issues

Build the order status workflow

Create a workflow to handle order status inquiries:Trigger: Tag added — “order-inquiry”

Look up the contact’s most recent order from synced data
If order found: Update the conversation with tracking information
If order not found: Ask for the order number or email address
If the order is delayed: Proactively apologize and offer a discount code for their next purchase
Add tag “order-status-resolved” when complete

Build the returns workflow

Create a workflow triggered by Tag Added — “return-request”:

Verify the order is within the return window
If eligible: Send return shipping label via email, provide return instructions
If not eligible: Explain the policy and offer alternatives (store credit, exchange)
Create a task for the fulfillment team to process the return
Send a follow-up in 7 days to confirm the return was received
Process the refund and notify the customer

Offering an exchange instead of a refund retains revenue. Train your bot to suggest exchanges first: “Would you like to exchange for a different size/color, or would you prefer a full refund?”

Build the abandoned cart recovery workflow

Create a workflow triggered by Cart Abandoned (via your e-commerce integration or custom event):Sequence:

Wait 1 hour after cart abandonment
Send SMS/chat: “Hi {{contact.first_name}}, I noticed you left some items in your cart. Can I help with any questions about [PRODUCT]?”
If reply received: Enable the bot to answer questions and guide them to checkout
Wait 24 hours: If no purchase, send a follow-up with a 10% discount code
Wait 3 days: Final reminder with urgency: “Your cart items are selling fast!”
If purchased: Remove from sequence, add tag “cart-recovered”

Do not over-message about abandoned carts. Three touches maximum. If the customer does not convert after three messages, they made a deliberate choice not to buy.

Build the product recommendation flow

When a customer asks for product help, the bot should:

Ask what they are shopping for (category, use case, occasion)
Ask about preferences (size, color, price range, material)
Recommend 2-3 products from the knowledge base that match
Include a direct link to each product page
Ask if they need help with anything else

Add product recommendation instructions to your prompt (included in the full prompt).

Full prompt

You are the customer support assistant for [STORE NAME], an online store selling [PRODUCT CATEGORY]. You help customers with order inquiries, returns, product questions, and shopping assistance.

IDENTITY AND TONE:
- You are [BOT NAME], a support assistant for [STORE NAME].
- Be friendly, helpful, and solution-oriented. E-commerce customers want fast answers.
- Keep messages concise — especially on SMS and chat. Use bullet points for clarity.
- Mirror the customer's urgency. If they seem frustrated, acknowledge it immediately.

ORDER STATUS INQUIRIES:
When a customer asks about their order:
1. Ask for their order number or the email address used for the purchase.
2. Look up the order and provide the current status.
3. If shipped: Provide the tracking number and estimated delivery date.
4. If processing: Let them know the expected ship date.
5. If delivered: Confirm delivery and ask if there are any issues.
6. If delayed: Apologize, explain the delay if possible, and offer a goodwill gesture.

RETURNS AND EXCHANGES:
When a customer wants to return or exchange:
1. Ask for their order number and which item they want to return.
2. Verify the order is within our [X]-day return window.
3. Ask for the reason (size, quality, not as expected, damaged, other).
4. If eligible: "I can start that return for you right away. Would you prefer a refund to your original payment method, store credit, or an exchange for a different size/color?"
5. If not eligible: "Unfortunately, this order is outside our [X]-day return window. However, I can offer [ALTERNATIVE — store credit, discount on next order, etc.]."
6. For damaged items: "I am sorry about that! Can you send a photo of the damage? I will expedite a replacement for you."

PRODUCT RECOMMENDATIONS:
When a customer asks for help choosing a product:
1. Ask what they are looking for (category, occasion, use case).
2. Ask about preferences (size, color, budget, material preferences).
3. Recommend 2-3 specific products with brief descriptions.
4. Include links to each product page.
5. Offer to answer any questions about the recommended products.

SHIPPING QUESTIONS:
- Standard shipping: [X] business days, $[X] (free over $[THRESHOLD])
- Express shipping: [X] business days, $[X]
- International shipping: [X] business days, $[X]
- All orders include tracking information sent via email.

DISCOUNT AND PROMO CODES:
- If a customer asks about current promotions, share any active codes from the knowledge base.
- If they have trouble applying a code, walk them through the steps.
- Never create or make up discount codes. Only share codes that are in the knowledge base.

BOUNDARIES:
- Never process refunds directly — collect the information and create a task for the support team.
- Never share other customers' order information.
- Never promise delivery dates that are not supported by the carrier's estimate.
- For payment issues (declined card, billing errors), direct them to contact their bank or our support email.
- For complex issues (fraud, chargebacks, bulk orders), escalate to a human: "Let me connect you with a specialist who can help with this."

CLOSING:
After resolving an issue: "Is there anything else I can help you with? We appreciate your business!"

Testing your e-commerce bot

Order status test

Provide an order number and ask for status. Verify the bot retrieves and displays correct tracking information.

Return request test (eligible)

Request a return for an order within the return window. Verify the bot walks through the return flow and offers refund, exchange, or store credit.

Return request test (ineligible)

Request a return for an order outside the return window. Verify the bot explains the policy and offers alternatives.

Damaged item test

Report a damaged item. Verify the bot asks for a photo and offers to expedite a replacement.

Product recommendation test

Ask for help choosing a product. Verify the bot asks qualifying questions and recommends relevant items with links.

Cart recovery test

Trigger an abandoned cart event. Verify the follow-up sequence fires at the correct intervals and the bot engages naturally when the customer replies.

Discount code test

Ask about current promotions. Verify the bot shares only valid, active codes from the knowledge base.

Escalation test

Report a complex issue (e.g., fraud). Verify the bot escalates to a human agent.

Optimization tips

Track top inquiry types. If 60% of conversations are order status, invest in making that flow seamless. If returns dominate, focus on reducing return reasons (better product descriptions, sizing guides).
Add proactive order updates. Instead of waiting for customers to ask, send proactive SMS updates when orders ship, arrive, or are delayed. This dramatically reduces inbound support volume.
Personalize product recommendations. Use purchase history and browsing data (stored in custom fields) to make recommendations more relevant. “Based on your last purchase of [PRODUCT], you might also like [PRODUCT].”
Optimize cart recovery timing. Test different intervals (30 min vs 1 hour vs 2 hours for the first message). The optimal timing varies by industry and price point.
Measure bot resolution rate by category. The bot may resolve 90% of order status inquiries but only 40% of return requests. Focus improvement efforts on the lowest-performing categories.
Add a satisfaction survey. After each bot interaction, send a quick 1-question survey: “How did I do? Reply 1-5.” Use the results to identify weak spots.

KPIs to measure success

KPI	Target	Where to find it
Bot resolution rate	70%+ of inquiries resolved without human escalation	Conversation AI dashboard
Average response time	Under 30 seconds	Conversation AI dashboard
Cart recovery rate	10–15% of abandoned carts recovered	E-commerce analytics + workflow tags
Return rate	Track and aim to reduce over time	E-commerce reporting
Customer satisfaction	4.5+ on post-interaction survey	Workflow survey results
Revenue from recommendations	Track purchases attributed to bot recommendations	Custom reporting
Support ticket volume	30%+ reduction after bot deployment	Comparison reporting

Next steps

After-hours support

Handle e-commerce support inquiries outside business hours.

Lead qualification

Qualify high-value wholesale or B2B inquiries.

Knowledge base

Expand your product and policy knowledge base for higher resolution rates.

Workflow automation

Build more sophisticated e-commerce automation workflows.

AI Overview

Agent Studio

Conversation AI

Voice AI

Content AI

Workflow AI

Use Cases

Advanced

AI Trust & Safety

Architecture overview

Step-by-step build

Full prompt

Testing your e-commerce bot

Optimization tips

KPIs to measure success

Next steps

After-hours support

Lead qualification

Knowledge base

Workflow automation

​Architecture overview

​Step-by-step build

​Full prompt

​Testing your e-commerce bot

​Optimization tips

​KPIs to measure success

​Next steps

After-hours support

Lead qualification

Knowledge base

Workflow automation

Architecture overview

Step-by-step build

Full prompt

Testing your e-commerce bot

Optimization tips

KPIs to measure success

Next steps