If you have ever put a long list of stops into a route planner and got back an order with obvious doubling back in it, you have probably run into a limit nobody advertises.
Most route planners stop genuinely optimising somewhere around 25 stops.
It is not a licence tier and it is not a bug. It comes from what is underneath.
Where the number comes from
Most consumer and small-business route planners are built on a commercial mapping API. Those APIs offer a waypoint optimisation option: hand it a start, an end, and a list of stops in between, and it returns a better order.
That option has a documented cap of about 25 intermediate waypoints per request.
So a planner built on it can optimise 25 stops properly. Past that, it has to do something else.
What apps do past the limit
Usually they chop your day into chunks and optimise each chunk on its own. Twenty-five stops here, twenty-five there, then the chunks are stitched back together.
Each chunk is genuinely well ordered. The problem is the joins. The end of chunk one and the start of chunk two were decided separately, so nothing ensures they are anywhere near each other. On a 100-stop day, those three or four joins are where the doubling back comes from.
Some apps cluster geographically first, which is better, but the clusters are still solved in isolation and the seams remain.
The result looks optimised, because most of it is. The waste is concentrated in a few transitions that are hard to spot on a list and obvious on a map.
Why it is not just about stop count
The bigger limitation is what a waypoint optimiser can express, which is almost nothing.
It solves one vehicle, visiting stops, minimising travel. That is the classic travelling salesman problem, and it has no vocabulary for:
- A delivery window on a particular stop
- How long you actually spend at each door
- How much a van can carry
- More than one driver
If your planner is built on that, those things cannot be part of the route. They can be displayed next to it, stored against a stop, shown as a note. They cannot influence the order, because the engine deciding the order has nowhere to put them.
This is why so many delivery apps have a time-window field that does not appear to do anything. It genuinely does not. It is a note.
What the alternative is called
The problem that includes those constraints has a name: the vehicle routing problem. It is a superset of the travelling salesman problem, and it is what delivery work actually is.
A vehicle routing solver takes stops, vehicles, windows, service times and capacities, and produces an assignment and an order for each vehicle. It is a harder problem and needs a different engine, which is why most apps do not do it.
How to tell which one you are using
You do not need the vendor to tell you. Two checks:
Put in more than 25 stops and look at the returned order on a map. If it is good in patches with a few unexplained long legs between them, you are looking at chunked optimisation.
Set a time window on a stop that is geographically inconvenient, then optimise. If the order does not change at all, the window is a note, not a constraint.
What RouteMate does
RouteMate runs a vehicle routing solver, so the whole list is sequenced in one pass, over real road distances rather than straight lines between pins.
Measured against production on 27 July 2026: 40 stops ordered in 138 milliseconds, and a full 498-stop route in 8.7 seconds.
Delivery windows, time on site, vehicle capacity and multiple drivers are part of that solve rather than fields beside it. When a window genuinely cannot be met, RouteMate names the stop and the reason instead of quietly reshuffling it, which matters more than it sounds: a route that silently breaks a promise you made to a customer is worse than one that tells you it cannot keep it.
Being fair about the comparison
Plenty of dedicated logistics platforms run proper vehicle routing solvers, and several competitors support windows and capacity perfectly well. The 25-stop ceiling is not a competitor problem, it is a characteristic of tools built on consumer mapping APIs.
The point is that the distinction exists and is rarely stated, so it is worth knowing which kind you are paying for.
Why it matters at your size
Under about 25 stops with no timed drops, this is academic. A waypoint optimiser will serve you well and you should not pay more for anything else.
It starts mattering when your day passes that count, when some drops carry windows, or when you run more than one vehicle. That is the point at which the engine underneath stops being an implementation detail and starts being the product.