How to plan a 20 stop delivery route
Twenty stops is an ordinary day for a courier or a service tech. It is also the point where the order stops being something you can see by looking at the map, and starts being worth real money every single day.
Short answer
On twenty addresses the order matters more than the roads. Sorting stops by neighborhood gets you most of the easy win. Computing the order across the whole set gets you the rest, and the rest is bigger than people expect.
Concretely, on the twenty stop day below: 95 miles for a carefully hand sorted list, 59 miles for a computed one.
What twenty stops actually costs
Take twenty addresses spread across a metro area and drive them in the order the jobs came in, which is how a list looks when it lands in your inbox. That is 181 miles.
Nobody actually works that way, so it is not a fair comparison and it is only worth mentioning as the size of the raw disorder. What a competent person does is sort the list by area first.
Grouping by neighborhood is half the job
Sorted by area, those same twenty stops come to 95 miles. That is a huge improvement over 181, and it took someone ten minutes with a map and some local knowledge.
This is the part people are good at. Humans are excellent at coarse spatial grouping: north side first, then downtown, then the east cluster. If you are doing this already, you have captured the biggest single win available.
The other half you cannot see
The computed order for the same twenty stops is 59 miles. Against the hand sorted list, that is 38 percent less driving.
The difference does not come from the grouping, which was fine. It comes from the seams: which cluster to start with, which corner of a cluster to enter at, which stop to leave for the drive between two areas, and whether one awkward address on the boundary belongs with the group before or the group after.
Those are exactly the decisions a person cannot hold in their head at twenty stops, and exactly the ones a distance matrix settles in a second.
Build the day in about five minutes
The whole thing is a paste, a glance and a button. The five minutes is mostly spent checking addresses, which you should do anyway.
What to fix before you drive
Three things are worth a look every single time, because each one is cheap to fix now and expensive to discover at 2pm:
- Outliers. One stop far from everything else will be flagged. Sometimes it is real, sometimes it is a geocoding mistake, and the two look identical in a list of text.
- Duplicates. Two orders for the same customer are one stop.
- Missing cities. A bare street address that resolved to the wrong town is the classic way a twenty mile detour enters a plan silently.
Fixed stops and time windows
Optimization is not always what you want, and a day rarely has twenty equally free stops. Common constraints:
- a morning appointment that cannot move;
- a pickup that has to happen before a dropoff;
- a customer who was told "early afternoon";
- finishing at the depot or at home.
Fix the start and the finish, keep the constrained stops where they belong, and let the order be computed around them. A plan that ignores real constraints is not a better plan, it is an unusable one.
Doing it every day
The savings only matter because they repeat. Six miles a day is a rounding error. Thirty six miles a day, five days a week, is roughly 7,500 miles a year of driving that produced no revenue: fuel, wear, and more importantly about an hour of driver time daily.
That is the real argument for spending five minutes on the order. Not the fuel, the hour.
FAQ
How many miles does a bad stop order cost?
On the twenty stop day above, 36 miles: 95 for a hand sorted list against 59 for a computed one. The gap grows with the number of stops.
Is sorting by neighborhood good enough?
It is most of the win and it is worth doing. It leaves the seams between clusters unsolved, which on twenty stops is still around a third of the remaining distance.
How long should planning twenty stops take?
About five minutes, nearly all of it checking addresses. If it takes longer, the addresses are going in one at a time and that is the thing to fix.
Can I keep some stops in a fixed position?
You should be able to fix your start and your finish at minimum. Stops with appointment times need to stay where they are, and no optimizer should move them.
What if two stops are at the same address?
Duplicates are merged on import. If they are genuinely two visits, keep them separate and expect the planner to treat the location as one point.
Does the plan account for traffic?
The order here is computed from road distances, not live traffic. Traffic is the navigation app's job while you drive, and it rarely changes which stop should come next.