Overview
Route planning looks simple until you try to do it well. Assign deliveries to drivers, sequence the stops to minimise driving time, respect the time windows each customer requires, stay within driver hours limits, match vehicle capacity to the load, account for traffic, and do all of this for a fleet of drivers with different start locations, different vehicle types, and different working patterns — before the first driver needs to depart. Done manually, this is hours of work that produces routes that are reasonable but rarely optimal. Done through software that applies combinatorial optimisation to the full problem simultaneously, it produces routes that are measurably better and takes minutes rather than hours.
The difference between manually planned and optimised routes is not marginal. Operations that switch from manual to algorithmic route planning consistently find reductions in total kilometres driven, improvements in time window compliance, increases in the number of stops per driver per day, and reductions in the overtime that poorly planned routes generate. At scale — fleets of tens or hundreds of vehicles making thousands of deliveries — these improvements translate directly to substantial cost reductions and service level improvements.
We build custom route optimisation software for logistics operations, delivery businesses, field service companies, and any organisation that plans routes for a fleet of vehicles on a recurring basis — from focused tools that optimise a single delivery operation to multi-depot, multi-fleet optimisation platforms serving complex logistics networks.
The Route Optimisation Problem
Route optimisation is computationally one of the hardest problems in operations research — a variant of the Vehicle Routing Problem (VRP), which is NP-hard, meaning that finding the mathematically optimal solution for real-world problem sizes is not computationally feasible. What is feasible — and what practical route optimisation software delivers — is solutions that are provably very close to optimal, produced in seconds or minutes using advanced heuristic and metaheuristic algorithms that explore the solution space efficiently.
The practical route optimisation problem has more constraints than the academic VRP. Real operations add:
Time windows. Each stop must be visited within a defined time window — the customer's stated availability, the business's opening hours, the appointment that was booked. Time window violations are not just operationally inconvenient — they directly affect customer satisfaction and service level agreements.
Vehicle capacity. Each vehicle has limited capacity — weight, volume, or both. Loads that exceed vehicle capacity require multiple trips or additional vehicles. Capacity constraints interact with delivery sequences because the weight and volume on the vehicle change as deliveries are made.
Driver hours. Commercial vehicle drivers are subject to legal working time limits. Routes that would require a driver to exceed their available hours are not valid solutions. Driver hours constraints interact with route length and with the time required at each stop.
Multiple depots. Operations with multiple warehouses or depots assign deliveries to the depot best positioned to serve them, considering both the distance from depot to delivery locations and the load balancing between depots.
Skill and equipment matching. Some deliveries require specific vehicle equipment — a tail lift, a refrigerated body, a crane — or specific driver certifications — a dangerous goods licence, a specialist equipment qualification. The optimiser must match deliveries to vehicles and drivers with the required capabilities.
Traffic and real-world travel times. Static distance-based routing produces routes that ignore the traffic conditions that determine actual travel times. Optimisation using time-dependent travel times — morning peak, midday, afternoon peak — produces routes that are realistic rather than theoretical.
Dynamic conditions. The plan that is optimal at 7am may not be optimal at 10am if a driver has been delayed, an order has been added, or a customer has changed their time window. Dynamic re-optimisation that updates routes in real time as conditions change keeps the plan achievable throughout the day.
What Route Optimisation Software Covers
Multi-stop route optimisation. The core function — taking a list of delivery stops with their time windows, the available vehicles and drivers with their constraints, and the depot location, and producing the optimal assignment of stops to vehicles and the optimal sequence of stops within each route. Optimisation runs in seconds for daily delivery operations with tens of vehicles and hundreds of stops, and within minutes for larger operations.
The optimisation objective is configurable — minimise total distance driven, minimise total time, minimise the number of vehicles required, maximise time window compliance, or a weighted combination that balances these objectives according to the operation's priorities. An operation that prioritises time window compliance over cost will produce different routes from one that prioritises cost over all else — the optimiser should reflect the operation's actual priorities, not a generic single-objective function.
Multi-depot optimisation. For operations with deliveries spread across multiple depots, multi-depot optimisation assigns each delivery to the most appropriate depot considering depot proximity, depot capacity, and the routing efficiency of combining deliveries from the same depot. Cross-depot balancing prevents situations where one depot is overloaded while another has spare capacity.
Territory planning and zone design. For operations with recurring delivery patterns — regular customers on defined delivery days — territory planning defines the geographic zones that each driver or depot serves, balancing workload across territories while minimising total distance. Zone boundaries that minimise cross-over between territories reduce the inefficiency that occurs when deliveries in the same area are split between multiple drivers.
Time window optimisation. Time windows that are too narrow constrain route efficiency by forcing sequences that are geographically inefficient in order to meet the timing. Time window optimisation identifies the time windows that, if relaxed slightly, would produce significant route efficiency improvements — providing the data that customer service teams need to negotiate time window adjustments that benefit both the customer and the operation.
For operations that offer customer-selectable time windows, time window generation from the optimiser — showing the available windows based on the current delivery load and the routing efficiency of adding a delivery to the day's routes — produces time windows that are achievable rather than aspirational.
Dynamic re-optimisation. The morning's plan is disrupted by the afternoon's reality. A driver calls in sick. A customer requests a change to their time window. An urgent delivery is added to the queue. A traffic incident blocks the planned route. Dynamic re-optimisation takes the current state of the operation — routes in progress, stops completed, current driver positions — and re-optimises the remaining work in real time, producing updated routes that incorporate the changed conditions.
Dynamic re-optimisation requires real-time driver position data from the driver mobile application and the ability to run optimisation continuously as conditions change — not a batch process that is run once in the morning and cannot respond to the day's events.
Capacity utilisation optimisation. Routes that are optimised purely for distance may produce vehicles that are underloaded — making stops that could be consolidated onto fewer vehicles. Capacity utilisation optimisation balances stop sequencing efficiency against vehicle fill rate, identifying opportunities to consolidate routes when vehicles are running below capacity and split routes when vehicles are overloaded.
What-if scenario planning. Before committing to operational changes — adding a new depot, changing the delivery day structure, taking on a new customer area — scenario planning tools allow the impact of proposed changes to be modelled against current demand data. The optimiser runs against the new configuration and produces the route plan that the changed configuration would generate, allowing the cost and service level implications to be assessed before the change is implemented.
Specialised Routing Problems
Multi-day routing. Deliveries that span multiple days — long-distance routes that require overnight stops, multi-day field service schedules, weekly delivery rounds — require routing logic that manages driver hours across multiple shifts, handles overnight locations, and balances the workload across the multi-day schedule.
Pickup and delivery routing. Operations that combine pickups and deliveries on the same route — collecting returns while making forward deliveries, picking up from suppliers and delivering to customers on the same vehicle — require routing logic that maintains the precedence constraint (the pickup must happen before the delivery for the same load) while optimising the combined route.
Service and field workforce routing. Field service operations — maintenance engineers, service technicians, installation crews — route technicians to customer appointments with the additional constraint that the technician must have the skills and tools required for each specific job. Field service routing optimises the assignment of jobs to technicians with the required qualifications alongside the geographic routing.
Periodic routing. Operations that deliver to the same customers on a recurring schedule — weekly delivery rounds, bi-weekly service visits — require periodic routing that produces stable, repeatable routes that customers and drivers can predict, while accommodating the demand variation that makes each period's routes slightly different.
Integration With Operational Systems
Route optimisation does not operate in isolation — it requires input data from operational systems and produces output that feeds downstream systems.
Order management systems. Delivery order data — addresses, time windows, order weights and volumes, special handling requirements — pulled from the OMS to populate the optimisation input. Updated delivery status from the driver mobile application pushed back to the OMS.
Warehouse management systems. Pick and pack completion status from the WMS confirms which orders are ready for loading before routes are dispatched. Load sequence optimisation — sequencing the loading of the vehicle so that the last delivery is loaded last — requires the route sequence before loading begins, creating a tight integration between route planning and warehouse loading workflow.
Fleet management systems. Vehicle availability, current vehicle locations, and driver hours remaining from the fleet management system feed the optimisation as constraints — ensuring that routes are built around what is actually available rather than the nominal fleet.
Driver mobile application. Stop completions, failed delivery recording, and real-time driver position from the driver app feed dynamic re-optimisation throughout the day.
Traffic data providers. Real-time and historical traffic data from HERE Technologies, Google Maps Platform, or TomTom feeds travel time estimation for both static optimisation and dynamic re-optimisation.
Exact Online / AFAS. Route cost data — driver hours, fuel cost, vehicle cost per kilometre — fed to the financial system for route profitability analysis and cost centre reporting.
Technologies Used
- Rust — route optimisation engine, combinatorial algorithm implementation, high-performance constraint evaluation
- C# / ASP.NET Core — optimisation service API, order management and fleet system integration, complex routing business logic
- React / Next.js — route planning interface, map visualisation, scenario planning tools, operational reporting
- TypeScript — type-safe frontend and API code throughout
- SQL (PostgreSQL, MySQL) — delivery order data, route history, optimisation run records, performance analytics
- Redis — dynamic re-optimisation state, real-time position data, optimisation job queuing
- HERE Technologies / Google Maps Platform — geocoding, routing, real-time traffic, map display
- React Native / PWA — driver mobile application with real-time route updates
- Shopify / WooCommerce APIs — order management system integration
- REST / Webhooks — fleet management, WMS, and OMS integration
- SMTP / SMS / push notifications — route dispatch notifications, dynamic update alerts
Measuring Optimisation Performance
Route optimisation performance is measurable — and measuring it is essential for demonstrating the value of the investment and for identifying where further improvements are possible.
Distance and time per stop. The primary efficiency metric — total kilometres driven divided by stops completed. Optimised routing should produce a measurable reduction in this metric compared to manually planned routing, holding the delivery area and stop density constant.
Time window compliance rate. The percentage of deliveries completed within the promised time window. Optimised routing that correctly models time windows should produce higher compliance rates than manual routing, which often sacrifices time window accuracy for geographic efficiency.
Vehicle utilisation. The average fill rate of vehicles — the proportion of vehicle capacity used on each route. Higher utilisation means more deliveries per vehicle per day and lower cost per delivery.
Failed first delivery rate. The percentage of delivery attempts that fail because the customer is not available or the address cannot be found. While this metric is primarily driven by customer communication and address data quality, routing that respects stated time windows and generates accurate arrival estimates contributes to lower failure rates.
Driver overtime. Routes that run over their planned time generate overtime costs. Optimised routing that correctly models travel times and stop durations produces routes that complete within the planned working time more consistently than manually planned routes.
Optimisation That Pays for Itself
Route optimisation software has a clear and quantifiable return on investment — fuel cost savings, reduced vehicle wear, reduced overtime, increased stops per driver, and improved service levels that reduce customer attrition. For operations above a certain scale, the savings from optimised routing exceed the cost of the software by a significant margin.
The breakeven analysis depends on operation size, current routing efficiency, and fuel and labour costs — but for most fleet operations above twenty vehicles making regular deliveries, the investment in route optimisation software is recovered within months and continues to produce savings indefinitely.