Route Optimization Software Development

What route optimization software actually has to solve

Putting pins on a map in a sensible order is the easy part. Real routing is a vehicle routing problem with constraints stacked on top: a bakery that must stock every cafe before 6 a.m., a medical courier whose lab pickups close at 4 p.m., a beverage distributor whose trucks hit their weight rating long before they run out of room. Tepia starts every routing engagement by writing those rules down, because a solver can only respect the constraints it is given.

The inputs matter as much as the math. Each stop needs a clean geocode, a realistic service time and a time window the customer actually honors. A stop that takes 4 minutes on paper but 20 minutes in practice, because the receiver makes drivers wait at a back door, will wreck every route that contains it. Fixing service time estimates often improves route quality more than any change to the algorithm.

Then there is the objective. Some operators want the fewest vehicles on the road, some want the fewest miles, some want balanced workloads so no driver finishes at noon while another works until dark, and many want all three in a set order of priority. Tepia makes that choice explicit in Discovery so dispatchers understand why the software built a route the way it did.

Constraints a custom routing engine should handle

Packaged route planners handle the basics well. Custom software earns its place when your operation has rules they cannot express. These are the constraints Tepia models most often:

Tepia encodes each constraint as hard, never broken, or soft, broken only with a penalty, and gives dispatchers a plain reason when a stop cannot fit. That transparency is what makes planners trust the output instead of rewriting it by hand each morning.

Choosing the routing engine and map data

You rarely need to invent a solver. Tepia usually builds on a proven engine and spends the effort on data, rules and the experience around it. The right choice depends on volume, vehicle type and the degree of control you need:

• Hosted optimization APIs such as Google’s Route Optimization API or HERE Tour Planning, which are quick to stand up and include traffic aware travel times.
• Open source routing with OSRM or Valhalla on OpenStreetMap data, hosted on AWS, Azure or GCP, when you plan very large volumes or need custom road rules.
• Open source solvers fed with your own travel time matrix, when your constraints are unusual enough that a hosted API cannot represent them.
• Truck attribute routing for vehicles where height, weight, axle count and hazmat class change which roads are legal.

Geocoding gets its own workstream. Rural addresses, new subdivisions, loading docks behind a building and gated communities all produce pins in the wrong place. Tepia builds a correction loop so a driver can drop the true delivery point once and every future route uses it. Our maps and geolocation team handles the map layer and travel time matrices.

Live reoptimization, ETAs and dispatcher control

A morning plan survives until the first cancellation. Good route optimization software reacts during the day: an urgent order slots into the best route that still has capacity, a failed stop moves to tomorrow, and a broken down truck’s remaining stops get split across nearby drivers. Tepia builds this as a dispatcher decision, with the software proposing changes and a person approving them, so drivers are not surprised by a route that reshuffles under them.

Dispatchers also need to lock what the solver should not touch, such as a stop a customer was promised first or a driver’s school pickup at 3 p.m. Tepia adds pinning, drag and drop editing with instant feasibility checks, and a log of every override. ETAs go out through Twilio text messages, recalculated from live GPS so customers see an honest arrival time rather than the one printed at 5 a.m.

Finally, measure the gap between planned and actual. Tepia stores planned miles, stop order and arrival times next to what really happened, which shows where service times are wrong and which customers consistently delay drivers. That feedback is how routes improve month after month. For prediction models layered on top, see AI route planning and ETA prediction.

How Tepia approaches route optimization software

Tepia applies its six phase process to routing work, with a US based project manager accountable throughout.

1. Discovery: a review of three to six months of historical orders and GPS traces, interviews with dispatchers and drivers, and an audit of your order system and telematics. Deliverables are an Investigation Summary with a baseline of current miles and on time performance, an Interview Summary and User Stories such as a dispatcher inserting a rush order at 11 a.m.
2. Design: wireframes for the planning board, route map, driver stop list and customer ETA page, plus a written constraint catalog your team signs off.
3. Development and Testing: the solver is replayed against historical days before any driver sees it. Alpha runs in shadow mode beside your current plan; Beta puts one depot on live routes, with functional, user acceptance and load testing.
4. Training: hands on sessions for planners on overrides and exceptions, and short driver sessions on the mobile app.
5. Launch: migration of customer locations, time windows and service times, then a depot by depot transition from the old planning method.
6. Support: tuning penalties and service times, bug fixes and new constraints as your network grows.

Typically Discovery takes 1 to 3 months, development and QA 2 to 6 months, and launch about a month. Details are on our process page.

Buy a route planner, call an API or build custom

If you run a handful of vans with simple windows and no integration needs, a packaged route planner is usually fine, and Tepia will say so. A middle path is common: keep your order or ERP system, call a hosted optimization API, and build a dispatcher board and driver app that fit your process.

A fully custom build makes sense when:

• Your constraints, such as paired pickups, multi day routes or mixed vehicle classes, exceed what packaged tools can express.
• Routing must sit inside an existing platform your customers already use.
• You plan thousands of stops a day and need control over performance and hosting.
• Routing is part of the product you sell, not just an internal tool.

Related reading: last mile delivery software, custom fleet management apps and custom delivery app development. When you are ready, talk to Tepia about your routes.

Frequently asked questions