Route and Schedule Optimisation: How to Plan Stops, Time Windows and Driver Shifts Together
Drafted with AI assistance, edited and fact-checked by Sean Flannery. See our editorial policy.
Route and schedule optimisation is the practice of deciding stop order, arrival time windows and driver shift capacity together, rather than one after another. It builds on the vehicle routing problem from operations research, and its goal is a plan that keeps expected travel time low and the risk of missing arrival windows low as well.
Picture a typical delivery morning. A planner sorts the stops into a tidy loop, then adds arrival times, then checks who is rostered and for how long.
Each step makes sense on its own. But when the times are added after the loop is fixed, the van can reach a customer well before their window opens, or after it has closed, with no easy way to reorder the day.
Planning all three as one decision puts the window inside the sequencing rather than leaving it as a check at the end. That is what keeps arrival promises realistic as the day fills up.
What route and schedule optimisation means, and why it works as one decision
There are three ingredients. The route is the order of stops. The schedule is when the vehicle should arrive at each one. Shift capacity is who is available to drive, in which vehicle, and for how many hours.
Plan them in sequence and each stage inherits the limits of the one before. The loop is already set when the windows arrive, and the windows are already promised when the roster is checked.
Plan them together and the optimiser weighs travel time against every window and every shift at once. A stop with an early window pulls its neighbours forward. A short shift caps how many stops a route can take on.
Here is a worked example on assumed inputs. One van, one morning shift, a depot on the west side of town. The east side customers all asked for early windows, while the west side stops are flexible.
Sorted by distance alone, the van clears the west side first and reaches the east side after the early windows have closed. Planned together, the optimiser sends the van east first, serves those windows, then works back west, and the shift still finishes on time.
It also helps to separate a few terms. Route planning sets out a route. Optimisation searches for the best combination that respects every window, shift and vehicle limit. Tracking shows how the day is going against that plan. The guide to how route planning differs from route optimisation covers that first distinction in more depth.
SuperPharmacy's case study shows joint planning at work in a busy daily operation. The online pharmacy has delivered prescription medications and health essentials to customers since 2000, and its delivery team was spending up to one hour every day planning routes by hand. A new urgent prescription during the day meant regenerating the entire route from scratch.
With Locate2u's route optimisation, routes are now generated automatically from delivery locations, time windows and driver availability. The hour of planning now takes 10 minutes, a cut in route planning time of 83% for this one pharmacy in Australia.
New prescriptions can be slotted into existing routes on the fly without disrupting other deliveries, which let SuperPharmacy offer same-day prescription delivery as a core service. In the team's words: "Before Locate2u, we had to generate new routes if we wanted to change anything. Now we can adjust the day's deliveries on the fly."
To test the idea on your own work, time how long tomorrow's plan takes to build and note how often a route is rebuilt after a single change.
From one van to a whole fleet: the routing models behind planning software
Planning software sits on a well-studied family of problems. Knowing which one matches your operation makes it much easier to judge whether a platform fits.
The travelling salesman problem asks for the best order to visit a set of stops with a single vehicle. The vehicle routing problem asks the same question for several vehicles, which adds a second decision: which vehicle takes which stops.
That second decision is where most of the work lies. Splitting stops across a fleet changes every route at once, so the problem grows quickly as vehicles and stops are added. The article on the maths behind multi-vehicle route planning walks through why.
Each variant of the vehicle routing problem matches a kind of operation:
A capacitated model fits any fleet where a vehicle can fill up before its shift ends. Multi-depot planning fits a business dispatching from two or more sites. Pickup and delivery fits returns, collections and transfers between locations. Time windows fit almost every customer-facing delivery.
Real operations combine them. A mixed fleet of vans and trucks needs a capacity limit per vehicle, plus a service-time profile per vehicle type, so a stop that needs a crane unload is never planned like a hand-unload parcel drop.
The table below defines the specialist terms used throughout this guide.
| Term | What it means in planning |
|---|---|
| Travelling salesman problem | Finding the best order to visit a set of stops with one vehicle |
| Vehicle routing problem | The same question for several vehicles, deciding which vehicle serves which stops and in what order |
| Routing with time windows | A routing problem where each stop must be reached inside an arrival window |
| Capacitated routing | Each vehicle has a load limit by weight, volume or item count |
| Multi-depot routing | Vehicles start and finish at more than one site |
| Pickup and delivery routing | Routes include collections as well as drops, with each collection served before its matching drop |
| Time window assignment | Choosing which arrival window to offer a customer as part of the plan, not as a fixed input |
| Service time | Time spent at a stop, from arriving near the address to driving away |
| Shift capacity | The working time each driver has available, including start, finish and breaks |
| Locked stop | A stop the optimiser must not move, such as a completed drop or a confirmed appointment |
| Re-optimisation | Re-planning only the remaining stops when something changes during the day |
| Planned versus actual | Comparing planned arrival and service times with the times recorded on the road |
| Empty running | Distance a vehicle drives with no load on board |
How choosing time windows alongside routes keeps arrival promises realistic
Most operators treat a time window as something the customer hands over. The research treats it as something the plan can choose.
A 2020 paper hosted by the Massachusetts Institute of Technology and accepted for publication in Transportation Science describes a routing model whose aim is "to simultaneously determine routes and time window assignments such that the expected travel time and the risk of violating the time windows are minimized."
That sentence is the core of this whole topic. The route and the promise are one decision.
In practice, it applies at the moment of booking. If the plan for Thursday already has a van working the north side in the morning, offering north side customers a morning window costs almost nothing. Offering one an afternoon slot means a second trip.
So what if your customers already pick their own slots at checkout? The same idea still works. Build the slot menu from the plan, showing only windows the routes for that area and day can realistically keep, and close a slot once its capacity is used.
Window width is a trade-off worth setting deliberately. Wide windows give the optimiser room to group nearby stops. Narrow windows suit perishable goods and appointments. Many operations run both, with narrow windows priced or reserved for the work that needs them.
Shifts belong in the same model. Enter each driver's start time, finish time and planned breaks as limits the optimiser must respect, so it never builds a route that only works if a shift runs long. Driving hours and break rules vary between countries, so check the current requirements with your national transport regulator before setting those limits.
For more on matching drivers to work across a roster, see the guide to driver routing and scheduling.
How accurate service times and clean inputs help each schedule hold
An optimiser can only keep the promises its inputs allow. Two inputs matter more than any other: how long each stop really takes, and where each stop really is.
Service time is everything between arriving near the address and pulling away. That includes finding a loading spot, walking to the door or dock, the handover itself and any unloading. A route can be the shortest on the map and still run late if every stop is planned at the same flat duration.
Stops differ. A crane unload at a building site, a signature at a pharmacy counter and a parcel at a front door each need their own service time. Set defaults by stop type, then refine them by location as records build up.
Location data deserves the same care. A pin dropped on the wrong side of a large site, a missing unit number or an outdated delivery entrance sends a capable driver searching. Clean addresses, accurate pins and short access notes, such as the gate to use or where to park, make every route easier to drive.
The best source of service times is the road itself. A simple planned-versus-actual loop turns each day's deliveries into better plans:
- Record typical access and loading time by location and time of day.
- Include that access time in each stop's planned service time.
- Sequence stops around likely loading access, not distance alone.
- Compare planned and actual arrival and service times every week.
- Update future service times from what the records show.
When comparing platforms, check that each completed stop records an arrival and completion timestamp you can export or report on. Without those, the loop above has nothing to learn from.
How same-day changes slot into a plan without rebuilding it
No plan survives the day unchanged. An extra order lands mid-morning, a customer cancels, a vehicle needs to come off the road. The goal is to absorb the change while drivers already on the road keep a stable sequence.
Start by deciding what is fixed. Completed stops are fixed. So are stops a driver is heading to now and any confirmed appointments. Mark these as locked, and let the optimiser move only what remains.
Set a cutoff for locking each route before it departs. Orders that arrive before the cutoff join the morning plan. Orders after it are slotted into routes already running, where a window and the vehicle's capacity allow.
Cancellations open gaps. A re-optimisation of the remaining stops can pull a later stop forward or move a nearby add-on job into the space.
If a vehicle becomes unavailable, re-plan its remaining stops across the vehicles still on the road. Where the fleet cannot cover every window, a revised arrival time sent early keeps customers informed and lets them plan around it.
All of this depends on seeing the day as it happens. Locate2u's real-time tracking shows where each vehicle is through the day, so a dispatcher can spot a route running behind early and move the next stop while there is still time to keep its window.
What road freight and workforce figures suggest about planning ahead
A few industry figures help put planning decisions in context. Each one points to a check worth making on your own routes.
Start with distance driven with nothing on board. Eurostat reports that at EU level, "around one fifth of total road freight vehicle-kilometres were travelled by empty vehicles (21.8%)" in 2025. UK Department for Transport road freight statistics put empty running at 30% of total loaded and empty vehicle kilometres travelled by GB-registered heavy goods vehicles in 2024.
These figures cover different vehicle groups and different years, so read them side by side as a sense of scale in road freight, not as a benchmark for a van fleet.
The decision they point to is a local one. Look at the leg after each route's last drop. If returns, collections or supplier pickups could sit on that leg, planning pickups and deliveries in the same model gives that distance a job to do.
City streets are set to get busier too. The December 2024 urban logistics white paper states that, without further intervention, the number of urban delivery vehicles "could increase by more than 60% globally by 2030," which the paper says would add to congestion.
For planning, that makes time-of-day travel variation worth building in now. Set windows by area and time band from your planned-versus-actual records, rather than from a single average travel time.
Finally, the workforce. The US Bureau of Labor Statistics projects employment of delivery truck drivers and driver/sales workers to "grow 7 percent from 2025 to 2035, much faster than the average for all occupations." The same page notes that most drivers receive instructions to go to a location at a particular time, and "it is up to them to determine the best route."
A plan that already sequences stops around realistic windows gives each driver a clear day from the first stop. Their local knowledge of loading bays, gate codes and quiet streets then feeds straight back into better service times and access notes.
How to start planning stops, windows and shifts together on your own routes
The safest way to begin is to test the approach on days you have already run. That way the result is measured against your own work, not a demo route.
- Pick a recent typical day and a recent busy day.
- Gather each day's real orders, windows, addresses, vehicles and shift times.
- Set service times by stop type, using any timestamps you already hold.
- Run the optimiser with windows, capacity and shifts as limits together.
- Compare planned finish times and window coverage with what happened.
- Choose locking rules and a cutoff for same-day orders before going live.
Then run the planned-versus-actual loop weekly for the first month. Tighten windows only where arrivals land consistently, and widen them where travel times swing through the day.
Two questions help when comparing platforms. Can it treat windows, vehicle capacity and shift limits as one planning problem? And can it re-plan the remaining stops mid-day while keeping completed and committed stops fixed?
The practical takeaway is simple: plan the order, the promise and the people as one decision, and let recorded service times sharpen it each week. To see how jobs flow from orders to drivers, look at Locate2u's dispatch planning, and when you are ready to replay one of your own days, book a call with the Locate2u team.
Frequently asked questions
Should you offer customers narrow or wide delivery windows?
Offer the narrowest window your plan can keep reliably: wider windows give the optimiser more freedom to group nearby stops, while narrow ones suit perishable or appointment-based work. A practical approach is to start wider in areas where travel times vary through the day, then tighten windows as planned-versus-actual records show where arrivals land consistently.
Does a mixed fleet of vans and trucks change how schedules are planned?
Yes, because each vehicle type needs its own capacity limit and its own service-time profile, so a stop that needs a crane unload is planned with a longer service time than a hand-unload parcel stop. Tag each order with the vehicle type it needs, so the optimiser only assigns it to vehicles that can carry and unload it.
How can you test route and schedule optimisation before changing your daily process?
Replay a recent, typical day: load its real orders, windows, shift times and vehicles, then compare the optimised plan's finish times and window coverage with what happened on that day. Running the same test on a busy day and a quiet day shows whether the plan holds across your normal range of volume.
What should happen to the plan when a vehicle becomes unavailable mid-shift?
Re-optimise only the remaining stops across the vehicles still on the road, keeping completed stops and locked commitments fixed so drivers already mid-route keep a stable sequence. Where the remaining fleet cannot cover every window, send affected customers a revised arrival time early or bring in extra on-demand capacity.
