Some will and some won't, and nobody can tell you the split for your callers. How to measure it on your own line in two weeks, and why telling them up front is the safer default.
Some will and some won't, and the split depends on the system and how it is set up. Don't rely on anyone's claim: measure it on your own line by comparing early hang-ups and booked calls before and after you switch it on, and tell callers up front what is answering.
Some will and some won't, and nobody can honestly give you the split for your callers without measuring it on your line. So don't build your setup on the hope that they won't notice. Assume they can tell, say so in the greeting, and judge the system on one number: does it book the job? You can measure that yourself in about two weeks, and the method is below.
Have the assistant say what it is in the greeting, something like "Thanks for calling Smith Plumbing, I'm the office assistant, I can get you booked or get a message to the tech." That costs about three seconds and removes the worst outcome: a customer who feels tricked after the fact. Disclosure and call-recording rules vary by state, so check yours before you go live.
These are the things you can check yourself on a test call, and the ones most likely to make a system sound like a machine:
Pull two weeks of call logs before you switch it on, then run it for two weeks and compare. You need three counts for each period: total answered calls, calls that hung up in the first 15 seconds, and calls that ended in a booking or a callback request.
Placeholder numbers to show the arithmetic (use yours):
In that example the early hang-up rate went up and you still booked 52 − 40 = 12 more jobs. If a job clears $300 gross profit, that's 12 × $300 = $3,600 for the two weeks. The hang-up rate is worth working on, but the booking count is the number that pays you.
If your real "after" bookings come in lower than "before", that's your answer too: change the greeting, the voice, or the hand-off rules and run another two weeks.
Whatever your numbers say, anyone who asks for a person should get one: a transfer during hours, or a specific callback time after hours. Make sure the system you pick can do both, and test it by asking for a human on your test call.
Call your current line from a cell phone five times at different hours and write down what happens. Then test-call any AI you're considering the same five times, using the checks above. Pick the one that books the job and hands off cleanly, not the one that sounds most human.
Ready to stop missing calls?
Bob answers every call, books every job, and handles your email — 24/7.
See Bob AI