"AI chatbot" gets sold as a single product, but what it actually does for a store ranges from "answers three FAQ questions" to "reads your order data and resolves a return." Before adding one, it's worth being specific about which of those you're paying for.
What a chatbot can realistically do
- Answer repeated questions instantly. Shipping times, return policy, sizing, order status � the questions that make up most support volume for most stores.
- Qualify a lead before a human gets involved. For higher-ticket or custom products, a chatbot can gather the basic details (budget, timeline, product interest) so a person isn't starting from zero.
- Work outside support hours. A visitor browsing at 2am gets an answer instead of a contact form and a wait.
What it can't do without real setup work
A chatbot is only as good as what it's connected to. Without integration into your actual order, inventory, and policy data, it either gives generic answers or makes things up � and a chatbot inventing a return policy is worse for trust than no chatbot at all. The setup work that actually matters:
- Connecting it to live order status, not a static FAQ file that goes stale the first time a policy changes.
- Defining clearly what it should escalate to a human, and making that handoff visible to the customer rather than looping them in circles.
- Testing it against your actual worst-case questions (damaged item, wrong address, refund dispute) before it goes live, not just the easy ones.
Signs your store is ready for one
A chatbot is a stronger fit when:
- Support volume is dominated by a small number of repeat questions (check your help desk's tag/category data if you have one).
- You get meaningful traffic outside your team's working hours.
- You have clear, current, written policies to connect it to � sizing, shipping, returns. If those don't exist yet, write them first; a chatbot can't answer questions your business hasn't answered internally.
Signs it's not the priority yet
- Your support volume is already low and mostly one-off, unusual questions � a chatbot has little to automate.
- Your policies change often and nobody owns keeping the chatbot's source data updated.
- The actual conversion problem is upstream of support � confusing product pages, unclear pricing, checkout friction � a chatbot won't fix a page that doesn't answer the buyer's question in the first place.
Chatbot vs. live chat � not the same decision
Live chat routes every conversation to a human; a chatbot handles what it can and escalates the rest. Most active stores end up running both � chatbot first, human fallback � rather than picking one exclusively. If your team is small, a chatbot without a genuine human escalation path can end up frustrating customers who need something it can't handle.
The honest limitation
No AI chatbot removes the need for a support process. It automates repetition inside a process you still have to define � policies, escalation rules, and what "resolved" means for your store. Skipping that step is the most common reason a chatbot project underdelivers.
Thinking about one for your store? Send us your brief and we'll scope the data integration and human-in-the-loop review it needs before launch.