How an AI Booking Assistant Recovered €22,344 in Bookings

The reception line rings several extensions at once. When everyone is occupied, the call ends. The caller moves on, often to a competitor, and no trace of them exists anywhere in the business. No ticket, no missed enquiry list, no line in a report.

This is what made the problem persistent rather than merely annoying. A report can only count things that happened. A call that never connected is a thing that did not happen, so it appeared in no dashboard, was never discussed in a meeting, and had never been quantified by anyone.

The team knew they missed calls. Nobody knew whether that meant five a week or fifty, and nobody knew what those calls were worth.

Every one of those callers had already decided to get in touch. The hardest and most expensive part of acquiring a customer, persuading them to make contact, was complete. The loss happened in the last few seconds of the journey, for want of someone free to pick up.

Compounding it, the same caller often tried once and never again. There was no follow-up because there was nothing to follow up from. The number was not captured anywhere the team could act on.

The first step was not automation, it was measurement: capturing every unanswered call so that for the first time the business could see the size of the problem. That alone changed the conversation, because the number was larger than anyone had guessed.

Only then did recovery follow. Each missed caller received a prompt, human-sounding follow-up inviting them to book, with reception alerted so a person could step in at any point. Because one incoming call rings several extensions simultaneously, the system had to recognise that as a single event and recover it once, rather than contacting the same person three times.

Nothing was written to the diary without a person, and the follow-up was designed to fail safely: where the system could not confidently answer a question about price or availability, it said nothing and passed the conversation to reception.

The full booked value across those 156 clients was 29,357 euro. The headline figure used here is 22,344, which strips out the single largest booking of 7,013 euro, because one fortunate client is not a business case. Strip out the largest booking of every single week and it still stands at 16,242 euro.

The median recovered client was worth 64 euro against a mean of 188, which confirms the result is carried by volume rather than by a handful of outliers. That distinction is the difference between a repeatable system and a lucky quarter.

The most useful outcome was not the revenue. It was that a category of loss which had been structurally invisible became a number the business now watches every week.

See the booking assistant →