Shoppers have questions and no one to ask
In a physical shop, a customer with a question asks someone. Online they either find the answer on the page or they leave, and most stores make them leave.
A chatbot answering from your own policies and product data closes that gap, and it does so at eleven at night when most browsing actually happens.
What it should answer
- Delivery — cost, timescale, cut-off, destinations
- Returns — window, condition, who pays
- Sizing and fit, from your own guides
- Compatibility — the highest-value question in many categories
- Order status, with an integration and a proper identity check
Compatibility and sizing are where most pre-purchase hesitation lives. A store that answers those instantly converts shoppers that a store with a contact form loses.
Where it should and should not appear
- Product pages: yes — that is where the questions arise
- Collection pages: yes
- Delivery and returns pages: yes
- Cart: carefully — helpful, but must not distract
- Checkout: no — nothing should interrupt a buyer with their card out
Never invent availability
A bot telling a customer something is in stock when it is not produces a cancelled order and a complaint. Stock answers must come from live data or not at all.
Without an integration, the correct answer is to direct them to the product page, which does show the truth.
Read what it cannot answer
The unanswered-question log from a store is unusually valuable, because it names the product information you are missing. “Will this fit a…” asked forty times is a product page that needs a compatibility section.
Adding that section helps the bot, helps human shoppers and helps your search visibility at the same time.