An ecommerce website chatbot for product and sizing questions
Shoppers hesitate on the product page: will it fit, is it machine washable, does it work with my model, when is the medium back. A website agent answers those questions from your catalog while the shopper is still deciding.
What does a chatbot on an ecommerce website help with before purchase?
It answers the questions that stop a shopper on the product page: size and fit, materials and care, compatibility, what is in the box, and when an item is expected back in stock, all from the catalog sheets, size charts and pages you upload. It can compare two products side by side and links to the right item. Questions it cannot answer go to your team with the transcript.
Worked example
Sizing questions on one jacket
Take one best-selling jacket. The questions shoppers ask in chat over a month might look like this (example counts):
| Example question | Count | Answer source |
|---|---|---|
| "I'm 5'10, 170 lb, medium or large?" | 38 | Fit notes and size chart you uploaded |
| "Is it warm enough for 20°F?" | 22 | Product description and fill details |
| "Can it go in the dryer?" | 15 | Care label sheet |
| "When is the large in olive back?" | 12 | Restock dates you published, if any |
The first row is the one to watch. If many shoppers ask the same sizing question, add the answer to the product page itself, and check whether returns for that item say "too small".
Setup checklist
Building a catalog the agent can answer from
- A product spreadsheet with one row per variant: size, color, material, care, compatibility, price.
- Size charts and fit notes per brand or product line, including how items fit compared with common brands if you know.
- Stock and restock information you are willing to share, kept current in a Google Sheet or connected through a webhook.
- Shipping and returns policies so the agent can answer "can I return it if it doesn't fit?"
Add it to any site with one line of embed code, or on Shopify by turning on the Botcadence chat widget as an app embed in your theme editor. It goes live in about 5 minutes once these files are uploaded.
Boundaries
Claims the agent should not make
The agent should not promise a fit, invent a restock date or make health, safety or performance claims that are not in your product copy. It does not take payments or apply discount codes that are not public. When a shopper asks about an existing order, it looks the order up if your Shopify store is connected and otherwise hands the chat to your support team.
Measuring it
What to look at after the first month
Track chats per product page, how often a chat ended with a product link click, and which questions went unanswered. Unanswered questions are the most useful output: each one is a gap in your product data that also affects shoppers who never open the chat.
See ecommerce AI answer a where-is-my-order question
Book a demo to map order status, return FAQs or cart follow-up onto Botcadence.
Common Questions
Ecommerce website chatbot questions
Only if you give it a current source, such as a Google Sheet you update or a custom webhook to your inventory. Otherwise it answers from the stock notes you uploaded and suggests the shopper check the product page.
Yes. Connect the store from your Botcadence dashboard, then switch on the chat widget under App embeds in your theme editor, with no code to paste. The connection also lets the agent look up orders; product and sizing answers still come from your product export, size charts and policy pages.
It can match the shopper's measurements against your size chart and fit notes and suggest a size, while saying that fit varies. It should not guarantee a fit you cannot stand behind.
Yes. It replies across 50+ languages from your English product data, which helps stores that ship abroad without translating every product page first.
Start with order questions or cart follow-up
Start with order status or cart follow-up, then add product recommendations and return FAQs once you can see what shoppers ask.
- Reader-first workflows
- Human handoff when needed
- Grounded in your content