Canned Responses and When Not to Use Them
Canned responses in Chat are simple by design — the judgment call is knowing when a static reply is the wrong tool.
Canned responses are table stakes for a live-chat tool — every product in this category has them, Chat included. The saved-reply CRUD is deliberately unglamorous: create a response, name it, store the text, reuse it. There's no scoring, no A/B variant testing, no approval workflow. The interesting question isn't whether the feature exists — it does — it's when to reach for it versus when not to.
What a canned response is good for
Canned responses earn their keep on questions that are genuinely repetitive and where the correct answer doesn't depend on the specifics of the conversation: "what are your hours," "here's the link to reset your password," "here's our refund policy." The visitor's situation doesn't change the answer, so a saved reply that gets the wording right once and reuses it every time is strictly better than an agent retyping the same sentence for the fortieth time.
Where it breaks down
The failure mode with canned responses isn't that they're wrong — it's that they're used on questions that only look repetitive on the surface. A visitor asking about a refund might be asking about your standard policy, or might be asking about a specific order that's already outside the standard window. A canned response answers the first case correctly and the second case in a way that reads as not having listened.
This is exactly the gap Chat's grounded AI reply suggestions are built to close. A canned response is static text chosen by an agent because the situation looked familiar. An AI-suggested reply is drafted per-conversation, pulled from your knowledge base with cited sources, and still requires an agent to review and send it — it's not automatic, and it's never sent to the visitor directly. The practical difference: canned responses are a shortcut for typing; AI suggestions are a shortcut for research. Use canned responses when the answer is genuinely the same every time. Reach for an AI-suggested draft when the situation needs a real answer written for it, just faster than typing one from scratch.
The judgment call
If you find yourself editing a canned response before sending it more often than not, that's a sign the question isn't actually repetitive enough for a static answer — it's a candidate for an AI-suggested draft instead, where the per-conversation context gets folded in automatically.