What Questions Should an Ecommerce Chatbot Answer?

Quick answer: An ecommerce chatbot should answer questions that have a definitive answer from a fixed source: sizing and fit, shipping timeframes by destination, return and exchange policy, stock availability, payment options, and product compatibility. It should not attempt judgment calls, complaints, refund disputes, policy exceptions, or emotionally charged situations, which belong with a human. The useful test is not “sales question or support question,” it is “does this have one correct answer I can ground in real data, or does it need judgment?”

The instinct is to make a chatbot answer everything. That is the fastest way to make it answer everything badly. The stores that get real value from one are ruthless about scope: they let the bot own the high-volume, definitive questions that repeat every day, and they route anything requiring judgment to a person. This guide draws the line clearly, and it is part of our ecommerce AI chatbot guide.

The real test: definitive answer vs judgment call

Forget “sales vs support.” Both contain questions a bot handles beautifully and questions it should never touch. The line that matters is whether the answer is fixed and sourced, or whether it depends on judgment.

• Definitive (let the bot answer). “Does this come in a medium?” “How long to the UAE?” “What is your return window?” “Is this in stock?” “Do you take Apple Pay?” “Is this compatible with X?” Each has one correct answer, retrievable from your size guide, shipping table, policy, or live inventory. These are the questions a human answering them is the most expensive way to do it.
• Judgment (route to a human). “Can you make an exception to your return policy for me?” “This arrived damaged and I am furious.” “Which of these two is genuinely better for my situation?” These need discretion, empathy, or authority the bot does not have. Forcing a bot to answer them is where trust breaks.

The questions worth prioritizing first

Not every definitive question carries the same weight. Start where volume and revenue overlap. “Where is my order” is consistently the single largest support category in ecommerce; Salesforce puts status-related questions at up to half of all inbound volume. And pre-purchase questions (fit, shipping, compatibility) are the ones that decide whether a hesitating buyer converts or leaves. In practice, the highest-value set to cover first is:

• Shipping: timeframes by destination, costs, and options.
• Sizing and fit: your size guide, and how items run (small, true, large).
• Returns and exchanges: window, condition, and how to start one.
• Stock and availability: what is in stock, and what to do when it is not.
• Product specifics: materials, compatibility, and the common objections buyers raise before buying.

Build these from your real support inbox, not assumptions, because the phrasing buyers actually use is rarely the phrasing you would guess. That audit-first approach is the core of our AI chatbot setup service.

How confident should the bot be before answering?

Not every definitive question deserves the same confidence bar. A useful way to think about it: the more a wrong answer costs, the higher the certainty you should require before the bot answers instead of escalating. Verifiable, low-risk questions (order tracking, standard shipping times) can clear a high automation bar because the data is checkable. Anything touching money (refunds, billing) should require more certainty and escalate sooner, because a wrong answer there is the kind that creates liability, as the Air Canada case showed. Softer questions (“which suits me better?”) are lower-stakes but also lower-confidence, so the bot should offer help while making the human option obvious.

Questions a chatbot should never answer alone

Some questions should trigger a handoff on sight, regardless of confidence:

• Complaints and anything emotional. A chatbot manages volume and predictability, not de-escalation. An angry customer needs a person, and a botched automated reply here is exactly what makes buyers abandon a store after one bad chatbot experience.
• Refund disputes and policy exceptions. These require authority and judgment the bot should not exercise.
• Anything it cannot source. New product launches, regional quirks, and one-off account situations often have no answer in the knowledge base. The bot should say so and hand off, not invent.
• Legal, safety, or medical claims. Never let a bot improvise here.

Designing those triggers is a conversation-flow and escalation task, and the difficult situations it hands off are a job for trained humans, not a failure of the bot.

Common mistakes

• Scoping too broad. A bot told to answer everything answers the important questions worse. Start narrow.
• Guessing the questions. FAQs written from assumption miss the real phrasing and the long tail. Pull them from the inbox.
• No “I do not know” path. A bot with no graceful way to say it lacks an answer will invent one.
• Same confidence bar for every question. Refund questions and sizing questions should not clear the same threshold.
• Letting it handle complaints. Emotional and dispute-heavy questions need a human from the first message.

Frequently asked questions

What questions should an ecommerce chatbot answer? Questions with a definitive answer from a fixed source: sizing and fit, shipping timeframes by destination, return and exchange policy, stock availability, payment options, and product compatibility. These recur daily and have one correct answer, so a bot can deliver them 24/7 without quality varying by who is working.

What should a chatbot not answer? Judgment calls: complaints, refund disputes, policy exceptions, emotionally charged situations, and anything it cannot source from its knowledge base. These need a human’s discretion, empathy, or authority. A chatbot that attempts them damages trust and can create liability if it invents a policy.

Should a chatbot answer product questions or just support questions? Both, as long as the answer is definitive. A chatbot handles pre-purchase questions like fit, shipping, and compatibility (which recover sales) and support questions like returns and stock equally well. The dividing line is not sales versus support; it is whether the answer is fixed and sourced or requires judgment.

How do I decide which questions to automate first? Start where volume and revenue overlap: order-status and shipping questions (the largest support category) and pre-purchase questions like sizing and compatibility (which decide conversions). Build the answers from your real support inbox rather than assumptions, so the bot recognizes the phrasing buyers actually use.

The question is not how many questions your chatbot can answer; it is which ones it should. Give it the definitive, high-volume questions that waste your team’s time, hold it to a higher bar the closer a question gets to money, and hand everything requiring judgment to a person. A chatbot scoped this tightly earns trust. One told to answer everything spends it.

Want a chatbot built from your real support inbox, scoped to the questions it should actually answer, with escalation designed in? That is our AI chatbot setup service, on your website and WhatsApp. Or book a free store audit and we will map which questions to automate first.

About the author

Mustajab Haider Bukhari is the founder of Organic Cart Studio, an ecommerce SEO, product copywriting, and customer communication agency specializing in Shopify and WooCommerce stores. Connect on LinkedIn.

This guide is educational and not legal advice; consult a qualified professional for compliance specific to your business and regions.