Delivery route optimization for small fleets
Delivery route optimization in plain terms: putting drops in order from the depot, sharing them between vans, reading road speeds and the limits to plan for.
Part of the guide Location intelligence automation
Delivery route optimization means putting the day's drops in the order that takes the least travel time from the depot and back, and, with more than one van, sharing the drops so each driver gets a compact area rather than a scatter across town. A small fleet gets most of the benefit from three habits: order the stops by road time, not by the order the orders came in; split the drops by area before ordering them; and look at live road speeds just before the vans leave.
This guide is for bakeries, florists, wholesalers, laundries, pharmacies and other businesses that run between one and a dozen vans. It explains what an optimizer actually decides, what it leaves to you, a made-up morning for a bakery in Lisbon, the mistakes that waste the effort, and how an AI employee can hand each driver an ordered list before they start. For the terms, see route planning in the glossary.
What a route optimizer decides
Given a start, a set of stops and the roads between them, a route planner tries orders of visit until the total travel time stops falling. With several vans it also decides which stops go to which van. The result is, for each stop, its place in the route, the distance and minutes from the stop before, and the totals for each route.
There are two ways to measure the roads. The better one follows the real road network for usual traffic. When that is not available, a planner can fall back to an estimate on straight lines with a typical speed, which gets the order roughly right but understates detours around rivers, rail lines and one-way systems. A good tool tells you which one you got; in xMap Flows every stop of an estimated route is marked as an estimate and the email says so.
What it does not decide for you
- Delivery time windows. A customer who only takes deliveries before 10:00 is not something the order of stops knows about. Put those stops in their own route, or first.
- Van capacity. The planner shares stops between vans by area, not by weight or volume. If one van fills up first, move stops by hand or split the list by size before routing.
- Time spent at each stop. Times are travel times only. Add your own allowance for parking, unloading and paperwork.
- Today's traffic. Routes are planned on usual traffic. A jam or a closure this morning needs a live check.
Reading live road speeds before the vans leave
Live traffic data does not count cars; it measures how fast traffic is moving now against how fast it moves when the road is clear. Two numbers matter around a depot or a cluster of stops: the current speed against the free-flow speed, and the jam factor, a score from 0 for free-flowing to 10 for standstill. A run of stops sitting in heavy traffic at 07:30 is a reason to send that van the other way round, starting at the far end.
A worked example, made up
A bakery in Lisbon delivers to 38 cafes and hotels each morning with three vans, leaving the bakery at 06:30. Until now the drivers split the list among themselves, and two of them crossed the same part of the city every day.
- The day's orders are shared between the three vans by area, so each van gets a compact part of the city: about 12 or 13 drops each.
- Each van's drops are ordered from the bakery and back by road, with a cap of 20 stops per route.
- Three hotels that need bread before 07:00 are pulled into a short first loop for the nearest van.
- At 06:15 the live speeds around the drops are checked. One cluster near the riverfront shows a high jam factor, so that van starts at the far end of its list.
The bakery's drivers now cover their areas without overlap, and the owner gets one email at 06:20 with all three routes. The numbers here are invented to show how the steps fit together.
Mistakes that waste the effort
- Optimizing the order but not the split. Three well-ordered routes that all cross the city are still worse than three compact areas.
- Planning on straight lines without knowing it. If the routing service was not used, the order may be fine but the minutes will be too low. Check the estimate flag.
- Forgetting the return. If vans must come back to reload, the last leg belongs in the total. Turn on the return to the start.
- Routing stops without a position. An order with a bad address cannot be placed. Fix addresses before the morning, not at the kerb.
- Overloading a route. Forty stops look efficient on screen and are impossible in a morning. Cap stops per route to what a driver can really do.
Each driver's route from an AI employee
In xMap Flows the morning routing can run on its own every day:
- Schedule: every delivery day at 06:00.
- New rows or My data: today's orders, each with an address and position.
- Split by territory: share the drops between the vans by area, each drop getting its driver.
- Route planner: one route for each driver, from the depot and back, by car, with a cap on stops.
- Traffic Flow: the live speed and jam factor around each drop, so a slow cluster stands out.
- Email or WhatsApp: each driver's stops in order, with the minutes between them and a link to follow.
The template New outlets, graded and routed to each rep shows the routing part working for a sales team, and Routes for the weekly surplus food pickup uses the same blocks for pickups. More on giving field work to AI employees is on the field operations page.
Questions
How many stops can one route hold?
In xMap Flows up to 50 stops per route. Most delivery drivers manage far fewer in a shift, so set a cap that matches a real morning.
Does it plan around delivery time windows?
No. It orders stops by travel time. Put time-critical stops in a separate early route, or first in the list.
Can live traffic tell me how many cars are on a road?
No. Live traffic data gives the current speed, the free-flow speed and a jam factor. Car counts by hour exist only for roads in the United Arab Emirates, through a regional dataset.
What if my vans start from different depots?
Every route in one flow starts from the same place you pick. For vans based at different depots, run one flow per depot, each with its own orders and start.