What is call analytics?

The basic layer comes free with any phone system. How many calls, when, how long, how many went unanswered.

That is enough to fix staffing. A business that discovers its busiest hour is 8am rather than 11am has learned something worth a rota change.

What it cannot tell you is why any of those calls happened. Duration is a poor proxy for value, and outcome codes are usually blank.

Adding transcripts changes the question you can ask. Counting how many callers asked about a service you stopped offering is a different kind of fact from counting calls.

Say an auto repair shop sets aside one hour on the first Monday of each month. The owner pulls last month's calls and writes four numbers on a whiteboard.

Total calls: 640. Unanswered: 71, with 38 of them between 7:30 and 9:00 when cars are being dropped off. Median call length: 1 minute 50 seconds. Then the top reasons people called, tallied from call summaries. Appointment requests lead at 212, is my car ready comes second at 164, and price questions follow at 97.

That second reason is the surprise. A quarter of all calls are customers asking for a status update, and each one pulls a service writer off the counter for two minutes. The shop starts texting customers when the car moves to the next stage. Next month the owner watches that tally to see whether it drops.

Average call length is the number most reports show, and it misleads. Ten calls of 90 seconds and one 25-minute warranty dispute average out at almost 3 minutes 40 seconds. No call that day looked anything like that. The median, which is the middle value when you sort them, stays at 90 seconds and describes your typical call honestly.

Small counts are the other trap. If 8 callers asked about tire rotation in March and 12 did in April, that's a 50% jump on paper and no news at all in practice. With counts under about 30, month-to-month swings are mostly chance. Look at the quarter before you change the sign out front.

Be careful comparing weeks with holidays, storms or a broken phone line in them. Note those on the whiteboard next to the number. Six months from now nobody will remember why February looked so strange, and the note stops you explaining a pattern that was only weather.

Every answered call produces a transcript and a short summary, so the reporting is about content rather than only duration.

You can count reasons. Ten callers asking for weekend appointments is the sort of finding that changes what you sell.

We publish no benchmark figures to compare yourself against, because we have none measured. Your own trend over a month is the number worth watching.