Shopify AI Chatbot Knowledge Base Checklist: What to Upload Before Launch
A source-pack checklist for Shopify teams that want grounded AI support before turning the widget loose on customers.
The fastest way to improve a Shopify AI chatbot is not to change the model. It is to upload the right store knowledge before launch. Shopify says quicker responses can increase conversion by up to 69%, but response speed only helps when answers match your store, policy, and order workflow.[1]
TL;DR
Before you launch a Shopify AI chatbot, upload product pages, return and exchange rules, shipping policy, order-status instructions, promo FAQ, warranty details, sizing guides, and approved support answers. Then define where AI stops: refunds outside policy, damaged goods, fraud suspicion, and angry complaints should hand off to a human.
Use this with our Shopify chatbot setup guide, Shopify AI chatbot page, and WISMO and returns workflow.
Why should the knowledge base come before the widget?
A Shopify chatbot that launches before the source pack is ready will answer fast and still create support debt. NRF reported that online returns were expected to reach 19.3% of online sales in 2025, so weak sizing, compatibility, delivery, or return guidance can become a margin problem.[2]
The widget is the easy part. The source pack determines whether the answer is useful. In our experience with document-grounded support workflows, bad sources create more risk than a basic interface ever does.
What is the minimum Shopify source pack?
The minimum source pack should cover product questions, returns, exchanges, shipping, order status, warranty, promo rules, and handoff. Shopify provides order status pages for customers to check order and delivery updates, so the chatbot should route WISMO questions into that self-service path rather than inventing status details.[3]
| Source | Questions it covers | Launch standard |
|---|---|---|
| Product pages | Fit, materials, ingredients, compatibility, bundles, availability. | Top products have current specs, variants, images, FAQ, and care notes. |
| Return and exchange policy | Eligibility, windows, final sale, condition, exchanges, store credit. | One current policy page, no old versions or conflicting macros. |
| Shipping and order status | Delivery windows, tracking links, carrier delays, WISMO. | Answer points to order status or documented tracking guidance. |
| Promo and checkout FAQ | Discount rules, bundles, free shipping thresholds, gift cards. | Campaign rules are dated and removed when the promo ends. |
| Support macros | Repeated replies, escalation paths, tone, must-say details. | Only approved current macros are included. |
Which Shopify sources should you upload first?
Upload sources in the same order your support queue creates cost: top-selling product pages, return policy, shipping policy, WISMO guidance, promo FAQ, then support macros. Shopify also warns merchants to edit theme code carefully, so knowledge setup should happen before the final embed step.[4]
Top-selling product pages
Start with products that drive the most traffic, questions, or returns. Include specs, sizing, compatibility, materials, ingredients, care instructions, and bundle notes.
Return, exchange, and refund policy
Use the current policy only. Add final-sale, damaged-item, international, exchange, and store-credit exceptions so the chatbot does not overpromise.
Shipping and delivery guidance
Include processing time, carrier windows, free shipping thresholds, local pickup, international rules, and what to do when tracking looks stalled.
Order-status instructions
Guide customers into Shopify order status or tracking guidance instead of asking the bot to invent an order state it cannot verify.
Promo and checkout FAQ
Add sale rules, bundles, discount stacking, gift-card limits, free shipping thresholds, and cutoff dates. Remove old promos quickly.
Approved human handoff rules
Document when the bot should stop: refund exceptions, damaged items, chargebacks, fraud suspicion, angry complaints, account access, and goodwill decisions.
What should you not upload?
More content is not automatically better. A Shopify chatbot gets worse when it sees stale, duplicate, private, or contradictory sources. Exclude old sale landing pages, draft policies, supplier-only docs, private customer data, unpublished pricing, legal notes that are not customer-facing, and support macros agents no longer use.
Practical rule: if you would not want a new support hire quoting the page to a customer, do not put it in the chatbot source pack.
If you want to test source quality before production, use a narrow crawl with Chat with Website or draft common questions with the AI FAQ Generator. Move only approved content into the production bot.
How do you QA the Shopify chatbot before launch?
The chatbot is ready only when it answers real support questions correctly and stops when source material is missing. Test product fit, return eligibility, WISMO, promo rules, warranty, and refund exceptions before the widget goes live.
| Test prompt | Pass condition | Fail signal |
|---|---|---|
| Can I return a sale item? | Cites current return policy and final-sale rule. | Says most stores allow returns. |
| Where is my order? | Routes to order status or tracking guidance. | Invents a shipping update. |
| Can I stack this discount? | Uses current promo FAQ or says it cannot verify. | Guesses based on generic promo rules. |
| My item arrived damaged. | Explains documented next step and hands off. | Approves refund or replacement alone. |
How should you maintain sources after launch?
Source maintenance should follow the retail calendar. Update the chatbot whenever products, policies, shipping cutoffs, promo rules, warranty coverage, or order-status instructions change. During sales and seasonal periods, review high-volume answers weekly because old promos and delivery promises go stale fast.
- Weekly in the first month: review no-answer questions, bad citations, handoff misses, and repeated product questions.
- Before major promotions: update discount rules, gift-card limits, free shipping thresholds, and cutoff dates.
- After policy changes: refresh return, exchange, warranty, shipping, and damaged-item guidance before traffic hits the bot.
- After new product drops: add product specs, sizing, compatibility, ingredients, and care instructions for the products customers ask about most.
What is the 7-day rollout plan?
A 7-day source-pack rollout works when you start with a narrow support surface and test real tickets before launch. Do not try to automate every Shopify issue on day one. Prove product, WISMO, returns, and promo coverage first.
- Day 1: export top tickets and tag product, WISMO, return, promo, warranty, and complaint intents.
- Day 2: collect product pages, policy pages, FAQ, order-status guidance, and support macros.
- Day 3: remove duplicate and stale sources before training.
- Day 4: add approved answers for returns, refunds, shipping, and promo edge cases.
- Day 5: run the QA matrix and fix content gaps.
- Day 6: test mobile widget placement and handoff behavior.
- Day 7: launch narrowly, then review no-answer questions before broader rollout.
Launch a Shopify Chatbot on Sources You Trust
Bring in your product pages, shipping rules, return policy, promo FAQ, and support docs. Then test DocMind on real Shopify questions before expanding automation.
Start Free with DocMindFAQ
What should I upload before launching a Shopify AI chatbot?
Upload product pages, return and exchange rules, shipping policy, order-status instructions, warranty details, promo rules, FAQ content, sizing guides, and approved support macros. Add human handoff rules for refunds, damaged items, fraud suspicion, angry complaints, chargebacks, and any exception that requires judgment.
Should a Shopify chatbot learn from every page on my store?
No. Start with high-value support sources first: product pages, policies, FAQ, shipping, returns, order tracking, warranty, and promo rules. Exclude stale, duplicate, draft, supplier-only, legal-only, campaign, and private customer-data pages that could confuse the assistant or expose information.
How do I test a Shopify chatbot knowledge base before launch?
Run real support questions through the chatbot before launch. Test product fit, shipping windows, return eligibility, promo rules, WISMO, refund exceptions, damaged item claims, and handoff. Every sensitive answer should match the approved source and should stop cleanly when source material is missing.
How often should Shopify chatbot sources be updated?
Update sources whenever product details, return windows, shipping cutoffs, promo rules, warranty coverage, or order-status instructions change. During sales, launches, and seasonal periods, review high-volume answers at least weekly because promo pages, delivery promises, and return exceptions change faster than normal FAQs.
What Shopify questions should still go to a human?
Human agents should handle refund exceptions, damaged-item review, missing-package investigation, fraud suspicion, chargebacks, angry complaints, goodwill credits, payment issues, and order changes the bot cannot verify. The chatbot can explain the documented policy, but it should not make discretionary business decisions.
About the author
Daniel Tang is the founder of DocMind. He studied Statistics at the University of Toronto, with coursework in machine learning and large language models, and previously worked at ByteDance as an AI Product Manager. This article was updated on April 14, 2026.
References
- Shopify App Store, Shopify Inbox, accessed April 14, 2026.apps.shopify.com
- National Retail Federation, 2025 Retail Returns Landscape, accessed April 14, 2026.nrf.com
- Shopify Help Center, Order status page, accessed April 14, 2026.help.shopify.com
- Shopify Help Center, Edit theme code, accessed April 14, 2026.help.shopify.com
- DocMind, Shopify AI chatbot, accessed April 14, 2026.docmind.com.au