Test Your Guide: Shop Knowledge That Keeps High-SKU Shopify Agents Honest

Look.

The fear is not that AI will talk to your customers.

The fear is that it will say something you would never say.

Wrong shipping cutoff. A product you do not stock. A return policy from 2023. A part number that sounds right and is not in Shopify.

That is why most high-SKU merchants treat storefront chat like a liability. On Reddit and in Shopify forums the same story shows up: the bot is generic, it does not know the catalog, and shoppers bounce the second it fakes an answer.

ShopGuide already reads your live catalog. Test your Guide is the quieter feature that makes that usable. It is the place you preview the agent, add shop knowledge, and catch bad answers before a paying customer does.

What merchants actually want to read

Talk to operators with 1,000 plus SKUs and they are not asking for another essay on the future of retail.

They want to know three things.

Can the agent find the right product when search and filters fail. Can I control what it says about my store. Can I try it myself before it goes live.

Shopify documents the filter wall in plain language. Collections over 5,000 products hide filters. A single filter can only show 100 values. Search stays literal. Shoppers type the problem they have. Your catalog is organized by the names you invented three years ago.

So they open chat. Or they leave.

If chat then hallucinates, you made the discovery problem worse. Test your Guide exists so that path is not a coin flip.

What Test your Guide actually does

In the ShopGuide admin it sits next to Dashboard and Chats. Two jobs, one screen.

First, preview. You talk to the same on-page agent your shoppers will see. Same catalog connection. Same settings. You can ask the ugly questions on purpose: compatibility, stock, "do you have X," gift constraints, policy edge cases.

Second, shop knowledge. You add the store facts that do not live cleanly on a product record. Free shipping thresholds. Return windows. Brand rules. How you sell sets versus singles. What the agent should say when something is out of stock.

That knowledge sits on top of live Shopify data. The agent still queries the Catalog API for titles, variants, metafields, inventory, and price. Shop knowledge fills the gaps the product record never held.

Chats then feed the loop. Recurring unanswered questions show up as knowledge gaps. Those are not vibes. They are the questions your catalog and your current shop knowledge did not cover.

Shop knowledge versus a static FAQ

A FAQ page is a document. Shop knowledge is input the agent can use while it is looking at current stock.

The useful combination is the last two together. Catalog for what you sell. Shop knowledge for how you operate. Preview so you see both before the storefront does.

You do not need a full knowledge-base rebuild to start. Add the five answers your team already types every day. Shipping. Returns. Lead times. Set quantities. What to offer when the exact SKU is gone.

The value that shows up for the merchant

Trust is the first value. You can sit in Test your Guide and break the agent on purpose. If it recommends a jacket you do not carry, you fix shop knowledge or the product data before a shopper sees it.

Tickets are the second. High-SKU queues are full of "does this fit," "when does this ship," and "what's the difference." Catalog-aware answers handle a lot of the product side. Shop knowledge handles the policy side so those do not bounce to a human by default.

Discovery is the third. Filters and keyword search still fail on big catalogs. An agent that knows both the live catalog and your rules can keep the session going instead of dead-ending on zero results.

None of that requires you to believe a marketing slide. You preview. You add the missing fact. You watch the next transcript.

Honest limits

Test your Guide does not replace clean product data. If critical specs only live in a messy description field, preview will show you the same sloppy answer a shopper would get. Fix the metafield or write the exception into shop knowledge.

It also does not make the agent a lawyer or a warehouse clerk. Custom quotes, damaged shipments, and account issues still need a person. Shop knowledge is for the repeatable store facts.

If you never open Chats, the knowledge gap list will not help you. The loop only works if someone reviews the questions the agent could not close.

And preview is only useful if you ask hard questions. "Hello" will not tell you if the 18V gasket answer is right.

Where this sits in agentic commerce

Harley Finkelstein has been blunt about the shift. Agentic shopping is merit-based. The thing that actually fits the need should surface, not the thing that bought the top search slot.

That only works if the agent is telling the truth about what you sell and how you sell it.

Merit dies the moment the agent invents a SKU or quotes the wrong cutoff. Shop knowledge is how a high-SKU merchant keeps the agent on their catalog and their rules. Test your Guide is how you check that before the traffic you already paid for has to find out the hard way.

Open the app. Ask the questions your support team is tired of answering. Add the two policies the transcripts keep missing. Then let the storefront agent talk.

Install ShopGuide and open Test your Guide

Frequently Asked Questions