Business AI adoption and competitor results: reading the numbers

Istat reports that 16.4% of businesses with at least 10 employees used an AI technology in 2025, up from 8.2% in 2024. This covers AI overall, not the phone receptionist market, and does not directly describe businesses with fewer than 10 employees.

It signals growing adoption, not a revenue forecast for a salon or practice. Your call costs and booking reliability remain more relevant measures for a purchase decision.

In its Fireman Hospitality Group case, Slang AI reports handling 27,824 calls and creating 2,415 voice reservations over 90 days. About 8% of reservations were outside typical business hours. The case concerns a restaurant group using an OpenTable integration.

This is a vendor source. It does not establish that all bookings were incremental or provide the net contribution attributable to AI here. A useful measure for your own trial is bookings outside business hours, counted separately from answered calls.

Calendly reports a 50% reduction in scheduling time for Find Your Fizz, a coaching business. This is a result in a vendor-published customer story about online booking.

It does not measure phone answering. It does show why simpler options deserve consideration: if the problem is exchanging messages to find a time, a booking link may already reduce the work.

An answered call, a request handled without staff and a completed appointment are different outcomes. To compare figures, ask for the period, call count, locations, starting position and counting method.

Treat revenue estimates based on multiplying every booking by an average spend carefully. They omit customers who would have booked anyway, cancellations and costs. A seller's case study is a useful example but is not an independent audit.

We do not yet have customer data to publish. A documented result will need a defined period, a baseline comparison and the customer's permission to publish it.

Useful measures include received and answered calls, accurate and completed appointments, requests passed to staff, correction time and full cost. Publishing the limits of a test makes its results more useful to buyers.