Drive time is the real capacity constraint
Branch leaders hire inspectors and crews to add capacity. A large share of the capacity they already own is sitting in a truck between two appointments that should never have been sequenced together.
the board nobody optimizes
Every foundation and waterproofing branch runs at least two schedules: an inspection board and a production board. They are usually managed by different people, in different tools, under different incentives. The inspection board is optimized for homeowner convenience. The production board is optimized for crew utilization. Neither is optimized for the thing that actually consumes the day, which is geography.
Book a 9am on the north edge of the territory and a 10:30 on the south edge and you have just spent ninety minutes of inspector capacity on asphalt. Do that four times a week across six inspectors and the branch has quietly lost the equivalent of an additional headcount — while the hiring requisition sits on the operations VP's desk.
the reschedule cascade
The second loss is worse because it is invisible in aggregate. Rain moves an exterior excavation. That crew now needs a different job today, which pulls a job scheduled for Thursday into Tuesday, which strands a material delivery, which pushes two homeowners who took time off work. Each hop is a phone call by a coordinator who is also answering new inquiries. By Wednesday, the board is a set of decisions nobody has time to make well.
A branch coordinator is not solving a routing problem in these moments. They are solving a triage problem with a phone and a memory of who is likely to be flexible. That is an extraordinary skill and an extraordinary bottleneck.
what an optimizer has to respect to be trusted
The naive version of this project fails predictably: an algorithm produces a mathematically superior route that ignores the crew that cannot do helical piers, the homeowner who can only do mornings, the permit that has not landed, and the fact that the same inspector should return to a repeat customer. Field schedules carry constraints that never made it into any system — and the fastest way to lose a branch manager is to hand them a plan that is obviously wrong on the first line.
So the sequence matters. Encode the hard constraints first: crew certification, equipment, permit status, weather sensitivity, homeowner windows, warranty callbacks that must be honored. Then optimize inside them. And propose — do not publish. A schedule that is recommended, explained, and one click from acceptance earns adoption. A schedule that overwrites the board earns a workaround.
the number to watch
Not miles saved. Billable hours per inspector-day and jobs completed per crew-week, held against callback rate so nobody buys throughput with quality. If the optimizer raises completed jobs while callbacks stay flat, it created capacity that would otherwise have cost you six hires and a year of recruiting.
sizing the drive-time loss branch by branch
Most operations VPs have never actually measured windshield time as a line item, because it does not appear on any invoice and no vendor bills for it. Pull the GPS or telematics data most trucks already generate and the number is usually startling: a meaningful share of a scheduled workday, often well over an hour per inspector or crew per day, spent driving between appointments that geography alone would have sequenced differently. Multiply that across a branch of six inspectors and two crews over a year, at fully loaded cost per hour, and the figure rivals the cost of the next hiring class the branch was about to justify to corporate. Presented that way, drive-time reduction stops being an operations nicety and becomes a capacity project with a clear return, which is the framing that actually gets budget approved at the VP level rather than languishing as an IT wish list item.
why acquisitions make this worse before it gets better
In a multi-branch network built through acquisition, territory boundaries are frequently inherited from the acquired company's original footprint rather than drawn around current crew locations and current demand density. Two adjacent branches, each running its own board independently, will routinely send an inspector forty minutes into what is functionally the other branch's territory because neither coordinator has visibility into the other's calendar, let alone authority to hand off the lead. Cross-branch capacity balancing — routing an inspection to whichever branch can actually reach it fastest, regardless of which branch generated the lead — is one of the highest-return, least-adopted capabilities in this vertical precisely because it requires the two coordinators, and often the two branch presidents, to agree to share demand instead of guarding it. That is an incentive design problem before it is a routing problem, and it needs to be solved at the compensation level before any optimizer can be trusted to act on it.
the ninety-day path to a trusted schedule
Spend the first month purely on measurement: instrument actual drive time, actual reschedule frequency, and actual reason codes for each reschedule, without changing anything about how the board is built. In the second month, encode the hard constraints branch by branch, in conversation with the coordinators who currently hold them in their heads, and run the optimizer in shadow mode — producing a recommended schedule that nobody has to act on, purely to compare against what the coordinator built manually. By the third month, the comparison itself tells you whether the tool is ready to be trusted with real proposals: if the shadow schedule consistently matches or beats the manual one on billable hours without violating a constraint the coordinator cares about, hand it the pen on new bookings while keeping humans in control of any reschedule involving an existing customer relationship. The endpoint is not an unmanned board. It is a coordinator who spends their day handling the genuine exceptions instead of doing arithmetic a machine should have done for them an hour earlier.