AI agents don't just recommend products — they answer questions. "Does this brand ship to Canada?" "What's their return policy?" "Do these shoes come in wide sizes?" When an agent can't answer those questions accurately, it either guesses (often wrong) or tells the shopper to check elsewhere. Either outcome breaks the handoff. This playbook makes your brand answerable.
Goal: Your brand's FAQ set is synced to Shopify's Knowledge Base and contains accurate, current answers to your top 10 most common shopper questions.
Time to complete: 30–45 minutes for setup; 15 minutes per update cycle thereafter.
Features used: Generate FAQs · My Store → Product Listing Editor
Step 1: Identify the Questions Agents Are Already Getting Wrong
Before you generate a FAQ set, it's worth knowing where your agents are currently failing. The fastest way to find out is to ask them directly — then compare the answers to your actual policies.
Open ChatGPT, Gemini or Perplexity and ask a set of common brand questions as if you were a shopper: "What's [your brand name]'s return policy?", "Does [brand] offer free shipping?", "Are [brand]'s products available in wide sizes?"
Note where the agent gives an accurate answer, where it hedges ("I'm not sure, check their website") and where it gives incorrect information.
These are the gaps your FAQ set needs to close — the questions where inaccurate or missing answers are already costing you shopper confidence.
Step 2: Build Your FAQ Set
Generate FAQs builds a store FAQ Knowledge Base that powers AI agent responses. Click + Add FAQ to open the import modal, then choose the tab that matches your source:
Go to Generate FAQs in the left navigation and click + Add FAQ.
Choose your source:
Generate — paste a URL to any page with FAQ content (e.g., your help center on Gorgias, Zendesk or Freshdesk), or upload a .txt, .md or .pdf file (up to 10 MB). AgentIQ extracts the Q&A pairs explicitly present in the source — it doesn't invent new questions.
CSV — upload a Q&A export from your support team. The file needs exactly two columns (Question and Answer), up to 100 rows per import. A Download Template link is provided.
Manual — author individual FAQ entries by hand, best for one-off gaps.
On the review screen (Generate imports only), check that every extracted answer is accurate and current — edit inline or deselect entries before importing. Agents will use this content verbatim.
Choose an import mode: Append adds the new FAQs alongside existing entries; Replace archives all current FAQs and replaces them with the new set (archived entries can be recovered).
Make sure the final set covers the specific gaps you identified in Step 1. If an important question is missing, add it via the Manual tab.
📌 Note: Quality matters more than quantity. Ten accurate, well-phrased answers to the questions shoppers actually ask are more valuable than 50 answers to questions nobody asks. Start with the questions your support team gets most often.
Step 3: Confirm Your FAQs Are Live in Shopify's Knowledge Base
The FAQ library is powered by Shopify's Knowledge Base — the layer that AI agents read when a shopper asks a brand question mid-conversation.
Imported FAQs are immediately active and available to AI agents — no additional step is required after the import completes.
If you make manual edits to existing FAQs afterwards (via the Edit pencil icon), click Sync to push the current library to the Shopify Knowledge Base so agents serve the most current answers.
Test it: go back to ChatGPT or Gemini and ask the same brand questions you tested in Step 1. Within 24–48 hours, the agents should be drawing on the updated FAQ data for your brand.
Step 4: Ensure Product-Specific Answers Are in the Listing
Product Listing page — Organization panel with metafields and description sections for product-specific Q&A data
The FAQ layer covers brand-level questions. Product-specific questions — "does this come in wide sizes?", "what material is this?", "is this compatible with X?" — need to be answered in the individual Product Listing. If the data isn't there, agents can't answer accurately even with a complete FAQ set.
Go to My Store and open a product that generates frequent Q&A in your support inbox.
Review the product description. Does it explicitly answer the questions shoppers most often ask about this product? If not, add those answers. Be specific: "Available in standard and wide widths (D and 2E)" not just "available in multiple widths".
Review the Metafields section. Fill in any structured data fields that correspond to common product questions: materials, dimensions, compatibility, care instructions, sizing details.
Repeat for your five to ten most-queried products.
📌 Note: The FAQ layer and the Product Listing work together. The FAQ handles brand questions; the listing handles product questions. An agent that has both sources can answer the full range of shopper questions with confidence — and a confident answer is what turns a conversation into a click.
Step 5: Keep the FAQ Set Current
A FAQ set that's out of date is worse than no FAQ set — it gives agents wrong answers to trust. Build a review cadence.
Any time a key brand policy changes (shipping thresholds, return windows, new markets, new sizing options), update the source and re-import the FAQ set — use Replace mode for a full refresh from your updated source of truth.
Before any major campaign (see Playbook 7), check whether the FAQ set needs seasonal context added.
After adding new product lines, check whether product-specific Q&A needs to be added to the new listings.
Related Articles