The Buffer Math Behind a Schedule That Doesn't Collapse by Wednesday

A schedule that assumes every job runs on time isn't a schedule, it's a wish list with timestamps. The fix isn't a better calendar tool. It's knowing the business's own variance.

Measure the spread, not the average

Pull the last 60 jobs and record estimated versus actual duration. The average will look fine. The spread won't. In most shops, the 90th-percentile job runs 45–70% longer than the estimate, and that tail is what destroys the afternoon.

Buffer at the day level, not the job level

Padding every job by fifteen minutes hides slack where nobody can use it. Instead, keep one deliberately unbooked block per tech per day, mid-afternoon. It absorbs the overrun, and on clean days it becomes the callback slot or a same-day add.

Protect the first job

The first appointment sets the tone for everything after it. Give it the tightest scope and the shortest drive. A late start compounds all day; a clean start buys the team an hour of margin by noon.

Make the reschedule rule explicit

Decide in advance who calls the customer when a job runs long, at what threshold, and with what offer. Written once, that rule prevents the daily judgment call that eats a dispatcher's afternoon.