A stage-by-stage map of an enquiry, from first contact to a confirmed booking, showing which steps are safe to automate first.
Key takeaways
Most restaurants and hotels lose the most ground in the first few minutes after an enquiry arrives: the phone rings during service, or a message sits in an inbox until someone has a free moment. This is also the lowest-risk place to introduce AI, because the tool is answering and booking, not making an exception to policy.
A well-scoped first-response pilot typically covers acknowledging the enquiry, answering the basic availability question, and confirming a straightforward booking. None of that requires the assistant to waive a fee or bend a policy.
Handling a group booking or a rate exception is where operators get cautious, and reasonably so. The setups that work well are ones where AI confirms standard availability and party size, while a manager handles anything outside the usual pattern, a large group, a special rate, a complex dietary request.
Treat any tool that proposes to waive a policy or set a custom rate on its own as out of scope for now. The goal is faster, better-organised booking for your team, not a faster exception with no one accountable for it.
Once an enquiry goes quiet, a meaningful share of lost bookings sits in follow-up that simply never happens because staff are focused on the guests already in front of them. AI can draft a check-in message and flag enquiries worth a second look, which staff review before anything goes out.
This is often where operators see the fastest, least controversial wins, because the volume is high, the individual messages are low-stakes, and the review step is quick to fold into an existing routine.
Pick one enquiry channel, typically the phone line or website contact form, and run first response and confirmation end to end before expanding. This lets your team evaluate the tool on one well-understood flow rather than a change across every channel at once.
By day 90, most operators have enough evidence, response time, bookings confirmed, staff feedback, to make an informed case for expanding to group enquiries or a second location, rather than guessing from a vendor's pitch.
A 30-minute call is enough to tell you whether AI pays for itself here.
