Oat, charcoal and slate knit cardigans with a care tag beside the charcoal garment.
E Commerce

How to Test AI Chat Fabric and Care Answers on Shopify

Run a four-question check on one Shopify garment to catch AI chat answers that copy fabric or care instructions from the wrong color.

5 min read
Shopify AI chatapparelvariant care instructions

To test fabric and care answers in Shopify AI chat, choose one garment whose colors have different approved care instructions. Ask the same washing question about each color, then switch colors in the same conversation. The assistant should use the instructions for the exact variant, say when it cannot verify an answer, and keep any product recommendation tied to the correct option.

This test matters when care determines whether someone will buy. A shopper looking for a machine-washable sweater may pass on a dry-clean-only one, even when both colors share the same style name and product page.

Pick one garment with a real difference between variants

Start with a style where the approved material or care instructions vary by color or finish. Ask your merchandising team to identify the approved source for each option. In Shopify, a variant is a specific combination of options, such as a color and size. Shopify documents metafields for storing specialized information on variant detail pages. Check how your storefront and the assistant each get that information before testing. Shopify’s guide to product variants · Shopify’s guide to care instructions.

Use a test store or staging catalog for deliberate gaps and conflicts. Here is a fictional garment you can recreate. The care instructions are sample data for the exercise. Use the actual approved care label when testing a real garment.

  • Harbor Cardigan, Oat / M. Cotton. Machine wash cold. Available to buy.
  • Harbor Cardigan, Charcoal / M. Wool blend. Dry clean only. Available to buy.
  • Harbor Cardigan, Slate / M. Wool blend. Care instructions missing. Available to buy.
  • Product description. Introduces the style and lists the colors. For the conflict test, add a generic “machine washable” statement to this description in staging while keeping the approved variant instructions above.

Keep a copy of the approved source with the variant ID, market, date and storefront URL. Use the same market and inventory snapshot across tools. Treat a material name as one fact and washing instructions as another. The blend name alone cannot tell the shopper whether that particular garment goes in the washing machine.

Ask four questions and follow the recommendation

  1. Confirm an explicit match. Ask, “Can I machine wash the Oat cardigan in medium?” Expect a yes and the cold-water instruction. If chat shows a product, open it and confirm the destination selects Oat / M or gives the shopper an explicit way to choose it. A card that silently defaults to Charcoal fails the destination check.
  2. Switch colors in the same conversation. Follow with, “What about Charcoal, same size?” Expect dry clean only. The answer should retain medium and replace Oat with Charcoal. Repeat in a new chat to tell a conversation-context error from a bad catalog answer.
  3. Check an unknown. Ask, “Slate is wool blend too. Can I put it in the washer?” The care instruction is missing. A useful answer states that the assistant cannot verify it and directs the shopper to the care label or customer support. Copying either of the other variants’ instructions fails this test.
  4. Test a hard buying constraint. Start a fresh chat and ask, “I need the medium cardigan in Charcoal, and it has to be machine washable. Which one should I buy?” Expect the assistant to explain why Charcoal misses the requirement. It can offer the verified Oat option if it names the color change. Repeat with the generic product description enabled in staging. If the sources disagree and the assistant cannot resolve the conflict, it should say it cannot confirm the care instruction.

Compare Zipchat and Brambles.ai on the same garment

Zipchat documents a Materials & Care Guide that uses uploaded specifications for fabric and care questions. If you plan to supply a single PDF with several colors, label each row with the product and variant identifiers. Then run the color switch and missing-data questions to see whether the answer uses the right row.

Brambles.ai describes product discovery based on your configured catalog and product information. Its product discovery guide calls out variant details and whether a particular custom field or document is available to the integration. For this test, confirm where the approved care text is available in your setup and whether the shopper can open the exact option from the answer. Brambles.ai’s product discovery guide.

These are documented approaches. The four questions are a proposed test for your store. Use the same product, source facts and shopper wording on both tools and record which source each tool can read. If one tool needs a different source format, record that setup step with the result.

Decide what has to pass before launch

Log the question, the expected variant and care fact, the answer you saw, and the option selected after opening the product. Grade the answer and destination separately. An accurate sentence with a mismatched product card needs a different fix from an answer that copied the wrong care instruction.

Run the four questions in two separate sessions, one on mobile and one on desktop. Treat a wrong wash instruction, an unsupported assurance about Slate, or a product destination that silently switches the option as a reason to hold this use case. When a source or mapping changes, rerun all four questions. Expand to another style only when your team can explain the result and maintain the source data.

If you are evaluating Brambles.ai for your store, start with one garment where color changes the care instructions. Bring the approved variant records and these four questions to a pilot. Explore Brambles.ai for Brands and Retailers.

Explore Brambles.ai

Learn more about our AI-powered agentic commerce platform, agentic shopping, and shopping assistance solutions.

Explore More Insights

Discover more articles on AI, automation, and business innovation

View All Articles