Route optimization is the quickest and easiest way for a last-mile delivery business to get more efficient routes and cut costs. Despite its well-proven power to reduce mileage, many businesses are still stuck in their ways of manual route planning.
We believe that this is because the term route optimization is ill-understood as it can mean different things to different people. It conjures up an image of mathematicians sitting in a room full of formulas on whiteboards, which gives it an air of scary mystique.
The goal of this guide is to dispel the mystique and provide clarity on the topic of route optimization, through a practical lens of small and growing delivery businesses.
We'll also discuss the problems with conventional route optimization and introduce a new concept called Intelligent Route Optimization to address these.
What is route optimization?
Route optimization is the process of finding the shortest, most cost-effective routes between multiple destinations, while meeting real-world needs and business constraints.
If all you have is a dozen addresses and a single driver, a human route planner can do the job decently well. But as your business grows to a hundred deliveries per day with multiple drivers, the route planning puzzle gets extremely complex.
Route optimization software takes all the information needed to plan a round of deliveries — addresses, time windows, driver schedules, vehicle capacities, and more — and automatically creates highly efficient routes in a matter of seconds.
Why does this matter?
There are wider social and environmental benefits to route optimization, too. Route optimization can help relieve traffic congestion and reduce fossil fuel consumption.
A short history of route optimization
A complete guide isn’t complete without a history class, so we’d like to bring you on a brief journey of route optimization’s origins in academia. You are free to jump to the next section on route optimization in today’s world.
For as long as humans have been moving around the world, they’ve tried to find ways to do it more efficiently. Our ancestors built their first roads following animal tracks, to take advantage of efficient routes that had evolved over many years.
Starting with Leonhard Euler’s solution of the Königsberg bridge problem in 1736 (spoiler: there is no solution), mathematicians have tackled increasingly complex routing problems.
In the 20th century, the Traveling Salesman Problem (TSP) dominated: Given a list of cities, what is the shortest possible route between them that visits each city exactly once and returns to the starting point?
Mathematicians tried various approaches to tackling the problem throughout the 20th century— in the 1950s and 1960s the RAND corporation even offered prizes for solving it.
To this date, the TSP remains an unsolved problem – at least in the sense of finding the optimal route for problems of any size. Instead, many creative meta-heuristics have been developed to approximate a good-enough solution.
In present time
The Traveling Salesman Problem remains one of the most-studied challenges in computer science, along with variations that consider multiple routes like the Vehicle Routing Problem (VRP) and the Pickup and Delivery Problem (PDP).
And that’s just the start of the variations. When you consider constraints like time-windows or vehicle capacities, you now have the Vehicle Routing Problem with Time Windows (VRPTW) and the Capacitated Vehicle Routing Problem (CVRP). You’ll even find academic papers using terms that sound like you’re cursing: MDHFVRPTW – the Multi-Depot Heterogeneous Fleet Vehicle Routing Problem With Time Windows.
No wonder route optimization isn’t the most popular kid on the block!
One of the problems of academia is that it incentivizes scientists to publish ever-greater numbers of academic papers. And the best way to get published is to tackle a new variation of the problem, or improve on an existing algorithm by finding more optimal routes. This has led to a huge body of literature covering a plethora of variations with creative acronyms. Chasing the academic goal of finding the best route has led to numerous hyper-optimized, tailored algorithms.
But who cares if a route solution is 99% optimal versus 99.5% optimal? Especially if the 99.5% solution takes 22.6 years of compute time to calculate! That’s how much computation power was needed to set the world record on a TSP problem with 15,112 cities – with no obvious practical benefit.
Route optimization in the real world
Large corporations like UPS or FedEx have spent billions employing huge teams of academics to develop their own in-house algorithms trying to find the best routes. We wanted to bring the power of route optimization to small and medium businesses in an easy and accessible way.
First, we needed to tackle the branding problem so we coined the term Route Optimization back in 2012 to consolidate the various academic terms being used at the time. The few commercial routing softwares that existed before us called themselves “TSP and VRP solvers” or “Multi-stop Routing and Scheduling Optimizers” to name a few. We started marketing the term in earnest in 2015 and you can see on Google Trends that the term Route Optimization has increased 300% in popularity as compared to 2010. But even today, it is still a relatively little-known term.
Secondly, we developed an easy-to-use web application that anyone can learn in minutes – as we’ll see in the section Getting started with route optimization. Existing solutions in 2012 were very complicated for people without a background in Computer Science to get started with. Fast-forward a decade, now there are many good software options on the market – here’s a review of the Best Route Optimization Software in 2023.
And finally, the algorithms needed to account for real-world factors, because the real world is much more complex than the academic world (unfortunately). For one, academic papers describe routes on a blank piece of paper (called a Cartesian plane) whereas real delivery drivers need to follow the road network, with varying traffic patterns throughout the day and week.
What’s more, there are soft human considerations to route optimization that are also important to keep your drivers happy. We'll expand on this concept further in the section on Intelligent Route Optimization.
Algorithms vs humans in route optimization
Perils of manual route planning
Most small businesses that are just starting out will use Google Maps as a route planner or create delivery routes in Excel. While these are familiar tools and a good way to get started, it still requires a lot of manual work and is very time-consuming.
Stringing together a manual route planning process requires tribal knowledge. In Richard's situation, his main route planner was out sick one day, and everything just fell apart – nobody on his team knew how to plan routes.
In present time
Using commercial route optimization software, anyone can plan delivery routes. And when you do, you’ll not only save yourself the headache, you might also reclaim one to two hours of your life.
Even if a human being decides to spend two hours of their time creating routes, how good are they?
Route optimization algorithms have proven to create routes that are much more efficient than human route planners. In a 2017 academic study, researchers invited participants into a lab and presented them with a route planning puzzle on a piece of paper and gave them one hour to complete.
On the easiest variant of creating 4 routes to 31 stops (the above picture), subjects drew routes that were on average 9.8% longer than the optimal solution. Not too bad. On a slightly more complex variant for 6 routes and 39 stops, subjects were on average 20.5% longer than the best route.
You can see that human route planner performance rapidly degrades as the size and complexity of the route planning problem increases.
The other thing to note is that while 20.5% shorter routes may sound little, for a delivery business that delivers to a hundred customers a day, this can easily translate to thousands of dollars saved each month! For larger delivery operations with thousands of deliveries a day, the savings scale into the hundreds of thousands a year.
Lab-based puzzles on a sheet of paper are much simpler than real routing puzzles that delivery businesses face. As mentioned before, real delivery businesses need to keep the real road network into account (as opposed to drawing straight lines). Also, the only constraint in the above experiment is capacity, whereas in the real world, businesses often have to consider delivery time-windows, different vehicle types, and driver preferences to name a few.
Therefore, it is conservative to say that an algorithm can easily outperform human route planners in the real world by 20%~30%.
Case study: Spring Hope Food
We ran our own man vs machine experiment with the Spring Hope Food Drive. Landon spent 4 hours trying to organize his routes by hand using his usual method. He then uploaded the same data to Routific to optimize delivery routes – which took him only 3 minutes. The results were staggering: his drivers were able to get the same deliveries done driving 37% fewer miles and with 8 fewer cars.
Just like with chess, humans stand no chance against route optimization algorithms. With today’s technology, there's no good reason why one should still plan routes manually. Route optimization algorithms can take care of the route planning headache, do it in minutes as opposed to hours, and find routes that are 20%~40% shorter, all of which translate directly to your bottom line.
Intelligent Route Optimization
This obviously defeats the purpose of using a route optimization algorithm.
Below are the major reasons why some businesses are reluctant to adopt route optimization software, because they typically can do better than a dumb algorithm.
Some routing software may advertise that they incorporate traffic, but usually they just pull the traffic data from Google Maps and apply it in retrospect to the route solution. Admittedly, this is what we did back in 2015 ;-) But we learned that this isn’t good enough, as it can break constraints and result in inefficient route sequences.
Intelligent Route Optimization uses AI technology to predict traffic patterns that vary depending on time-of-day and day-of-week as a direct input to the algorithm, so that the algorithm can optimize with this in mind.
No more spaghetti routes
This was a large-scale operation that uses an expensive routing solution, yet employs eleven manual route planners to manually create clean clusters to keep their drivers happy.
Drivers also dislike getting off the highway for just a couple stops. They prefer clean clusters where they drive a long leg in the beginning and end, so they can focus their work in a dense area.
Spaghetti routes may be shorter mathematically in some cases, but they are not always better in the practical sense. Here’s an image that illustrates the difference:
You can also see the teal-coloured starting in the south and snaking up north, doing single deliveries along the way, passing by three colleagues along the way. Imagine the driver’s look on their face at the end of the day!
In a research led by Paolo Intini, he showed that drivers can drive 10% faster on roads they have driven on more than five times.
If on top of that you shave off a few minutes of finding the buildings, parking, and delivery location for a particular customer, your deliveries will be even more efficient – while reducing the error rates.
Happy drivers, fewer errors, lower cost-per-delivery, a more profitable business!
Intelligent Route Optimization incorporates drivers’ experience into the route solution intelligently, without sacrificing overall route efficiency.
If you pay your drivers by the number of packages, you want to balance the routes by number of deliveries. If you pay them based on the hours worked, you want to balance the routes by shift time. Even if your drivers are salaried, they would still prefer to be treated fairly.With smart auto-balancing everyone gets to go home early!
That’s the best case. In the worst case, the pin is put erroneously on the other side of town, leading to tons of frustration and wasted time, when the driver arrives there only to realize it’s the wrong address.
Make sure you have clean addresses, or choose routing software that has smart geocoding with fallbacks and highlights potentially risky addresses for manual verification.
Getting Started with Route Optimization
Our philosophy is that AI should not aim to replace human dispatchers. Instead, we liken Routific to an Iron Man suit – it gives the dispatcher superpowers! A single dispatcher can now manage a fleet of twenty drivers easily, whereas previously you needed to hire more people.
The accuracy of geocoding is vital. Sending a driver to the wrong location on the other side of town can be extremely frustrating and costly.
Good routing software will have very accurate geocoding processes and warn you about potentially faulty addresses to give you a quick way to review and fix those.
Most routing software will allow you to upload a spreadsheet of addresses. The addresses are then interpreted in a process called geocoding and plotted on a map.
The next step is to set up your delivery fleet. At a minimum, you need to specify the number of routes, the start locations, and the route shift times.
With Routific you can choose to spread the deliveries across a pre-defined number of routes, or let the algorithm figure out the fewest number of routes.
You can also specify a capacity on your vehicles to make sure routes are not overloaded – or assign a driver to a territory they're most familiar with.
With the data set up, you’re ready to create your first set of optimized routes!
Human brains are wired to be skeptical of algorithms and AI, especially when you’ve been doing it manually for years. That’s why we’ve created an intuitive and powerful route planning interface that makes it easy to inspect routes and make changes, should you feel the need.
Over time, most route planners learn to trust the algorithm and let it do most of the heavy lifting.
Choosing route optimization software
Capterra is a public reviews site where real users can rate software and leave feedback and testimonials. One thing to watch out for is Capterra’s list is by default sorted by “Sponsored” which means that whoever pays the most, gets shown on top. Make sure that you change it “Highest rated” to sort the list in a more useful way:
We showed that route optimization can reduce mileage by 20%-40% and how it is the single most impactful lever you can pull overnight to increase the gross margins of your delivery business. Aside from route optimization, it’s important to review your entire delivery management process that may benefit from automation as well.
Yet, many businesses stick to manual route planning. Those that do use route optimization software often still spend hours tweaking the routes to their likings. With Intelligent Route Optimization this is no longer necessary, it it considers driver happiness and route acceptability. Among other things, Intelligent Route Optimization is traffic-aware, creates non-overlapping routes, and provides smart recommendations as the system gets to know your drivers.
Frequently Asked Questions
No, Google Maps cannot optimize your route. Google Maps is created for consumers who travel from A to B. But when presented a list of addresses to visit, Google cannot give you the optimized sequence of stops.
Here's an article that describes the relationship between route optimization and Google Maps.
Given a list of addresses, drivers, and constraints, route optimization software will calculate the best routes possible and present them on a map, and in the case of Routific, on a timeline view as well. You can then inspect the routes and make manual adjustments as you see fit.
1. Automate route planning: the amount of time you spend manually planning routes every day can be better spent on growing your business.
2. Optimize routes: with route optimization you can find 20% to 40% shorter routes, which directly reduces your cost per delivery (which is made up of fuel costs and driver wages).
3. Reduce risk of tribal knowledge: most business owners that are just starting their delivery business often take on the burden of manual route planning. Over time, they acquire tribal knowledge on manual route planning. With route planning software, anyone can do it. And the business owner can finally take a vacation.
4. Tracking KPIs: with route planning software you get the benefit of visibility, how much you’re driving each day, your cost per deliveries, how much mileage each car has accumulated, which makes fleet management easier too.
5. Real-time driver tracking: most route optimization solutions come with a mobile app with GPS tracking, so you know where your drivers are in real-time.
Most route optimization software do not take traffic into account when creating the routes. They typically rely on average road speeds using mapping data from OpenStreetMaps. We discovered that the lack of traffic consideration leads to overoptimistic ETAs from OpenStreetMaps, which then leads to unrealistic routes your drivers cannot follow.
With Routific’s Intelligent Route Optimization, we do take traffic into account. We have trained 179 Machine Learning models across the world to predict traffic patterns, which we then incorporate into the route optimization algorithm. We have seen ETA accuracies improve significantly – sometimes as much as an hour on an 8-hour route.
Yes, there are many free route planning apps available, but they each have limitations.
We have written an in-depth review that lists the 7 Best Free Route Planners in 2023.
Routific’s route optimization software also includes delivery management functionality such as dispatching, live GPS tracking, and customer notifications.
Many other route optimization softwares also have this, though not all.
Routific’s Engine API is a standalone route optimization engine, accessible via a stateless REST API, with wrappers available for Node, Ruby, and Python.
We also have APIs available that allows you to connect to our full end-to-end delivery management platform.
Route planning is the entire process that starts from having a list of addresses, creating a multi-stop routes (either manually or use route optimization algorithms), inspecting and editing those routes, all the way to dispatching those routes to your drivers, ideally on their mobile app.
Route optimization is a part of the route planning process, which seeks to automate the process – so you spend less time route planning – and to find the most efficient route possible, thereby cutting mileage and fuel costs.
For more details, see Route Planning vs Route Optimization.