The red arrow showed up on a Monday. I opened the monthly fleet report with my coffee, and average speed was down 15 percent from the month before. A drop like that is enough for someone two levels above you to start asking questions you cannot answer yet, and by lunch I had an email asking exactly that.
I have stared at that arrow more times than I would like to admit. The number is almost never wrong. What it rarely does is explain itself, and the real explanation is usually messier than “the drivers got slow.”
Tom Peters spent the eighties telling managers to get out of the office and walk around, because that was where you saw what was actually going on. A dashboard gives you the what. If you react to the number before you know why it moved, you are guessing.
The Time I Panicked Over Nothing
I once came close to restructuring a whole route network over a speed drop that turned out to be construction on Highway 34. The interchange was down to one lane for two weeks, and every driver on the east corridor already knew about it and was handling it fine. A full day went into pulling telematics data and building comparison spreadsheets. I had a memo half written to the regional director about “emerging inefficiencies” when Ray, one of my dispatchers, walked over and asked if I knew they were repaving the interchange.
I did not know, because I had not asked anybody.
Now I make the phone calls before I open a spreadsheet. The people driving the routes usually know what changed weeks before the report shows it, and they will tell you if you ask something specific like “anything different on 34 lately?” instead of “why is everyone slow?”
One Number for the Whole Fleet
A slower average can mean drivers sat in more traffic. It can also mean they took it easy through a school zone that just reopened for the year, which is what you want them to do. The report shows the same figure for both, and you would handle them very differently.
The fleet-wide average is the bluntest number you have. It folds every route and every hour of the day into one figure that is technically accurate and close to useless for working out what went wrong.
Start by segmenting. Break it down by corridor first, then by shift. When I do this, most of the fleet is usually running normal and a handful of routes or time windows are dragging the average down.
The Usual Suspects
Traffic pattern shifts are the quiet ones. A new distribution center opens near one of your corridors, and the afternoon window that used to move fine is suddenly full of semis. When school starts, morning routes through residential streets lose seven minutes or so a day and do not get them back until June. These changes come on slowly enough that drivers adjust without mentioning it. Cities also retime signals or put in speed humps without any notice that would reach your desk. Your drivers noticed the week it happened, and the report caught up a month later. I have made a couple of bad calls in that month myself.
Construction season is worse than people expect.
One road closure does not just add a detour. Every other driver had the same rerouting idea, so the alternates fill up too, and a closure that should cost five minutes a trip ends up costing twenty because the narrow back roads were never built for that volume. Mapping apps play a part in this. Reading about how map providers collect and update road data explains why routing software can be days behind what is on the ground, and more than once I have had a driver call in from a line of orange cones on a road the map showed as clear.
The Uncomfortable Internal Stuff
Sometimes the drop has nothing to do with the roads. If you recently changed how drivers are rewarded, say from on-time bonuses to safety bonuses, expect the numbers to move. Drivers respond to what you measure and pay for, and that was the reason for the change, so do not panic when it works.
New hires drive slower.
My first month on a delivery route, I was the slowest guy on the board by a wide margin. I double-checked every turn and circled blocks looking for addresses the veterans could find half asleep. If you onboarded several new drivers last month, part of your drop is that same learning curve, and it recovers on its own. Pushing new people to match veteran pace before they are ready gets you accidents and quits. Either one costs more than a temporary dip on a report.
Route changes made with good intentions catch people off guard. I watched one operation lose real time after someone moved the lunch stops closer to restaurants. The stops were all in busy commercial areas, so drivers spent whatever they saved getting in and out of parking lots and crowded intersections. Lunch itself was no longer than before.
Give any route change a two-week check where you compare actual times to what was projected. Then be willing to roll it back if the times do not match the spreadsheet, which is the part most people skip because they already told their boss it was an improvement.
Do Not Forget the Trucks Themselves
Vehicle condition shows up in speed data long before it shows up as a breakdown. Underinflated tires make drivers ease off without thinking about it, and worn brakes make even experienced drivers more cautious in stop-and-go traffic. Neither one turns on a warning light. Put your maintenance backlog next to the speed drop. If several units are overdue, start there, because it is fully in your control and nobody has to change how they drive.
Weather gets forgotten because it seems too obvious to mention. I have watched managers spiral over an October-to-November decline that came down to rain and earlier sunsets, so compare against the same month last year before you compare against last month.
The Fix Everyone Reaches For First
The most common response to a speed drop is to lean on drivers to go faster, and it is the worst one. It works for a few weeks. After that come the incidents and the turnover, and those cost more than the minutes you were chasing. If your drivers serve customers who are waiting on them, there are better ways to take the heat off the clock. There is a post here about rental companies tracking their fleet without installing hardware, and the same thinking applies: a customer who can watch the driver get closer on a map tends to call dispatch less about when.
A 15 percent drop is big enough to investigate and small enough that it almost never has a single cause. In my experience it is a few smaller things stacked together. Fix the ones you control, and tell your boss plainly which ones you do not.
Before you build the spreadsheet, call a few drivers. If you want something to check their story against afterward, live location for your team’s phones shows where each person is while they are on a trip, and the trip replay plays that trip back with the clock running, so you can see where someone sat still on 34.