Sales territory management is a balancing act. Maintaining equity between salespeople, maximizing profitability, and minimizing costs are important and interconnected factors. There is no one-size-fits-all solution, and if the balance is off, sales management is likely to leave someone within their organization unhappy or leave money on the table.

Using spatial analytics and data science to optimize territories based on company priorities can help sales managers to eliminate guesswork and create well-reasoned and data-driven territories. Using the example of a medical device sales company in Illinois, we dove into the methodology around building optimal sales territories.

Identify Priorities, Build Baseline Solutions

For our example we are working to optimize sales territories for a company that has 40 sales reps targeting 925 medical centers across the state of Illinois, which the reps visit each month. In order to create a baseline territory solution, we first need to understand this company’s priorities.

The medical sales company is looking to minimize sales rep travel to cut costs and increase productivity. They also want more evenly valued accounts across reps - allowing for a fairer commission plan across the board.

Visualize your data

Sales Rep Home Locations

As a first step, we visualized the home location of each sales rep, as in the map above, and then mapped each of our 925 medical centers, styled to show both the type of facility and their patient capacity.

Medical Centers

By looking at this initial data, we can understand how our reps and medical centers are distributed, with large groupings of both centered in urban areas, particularly in Chicago.

As one of the company’s stated priorities is evenly valued accounts across rep territories, we also need to assign value to each medical center. Patient capacity is one key indicator of value, which may pose a challenge, as it is not evenly distributed across medical centers.

Patient Capacity Per Medical Center

That said, patient capacity is not the only measure of value. To get a clearer picture of the demand value at each center, we looked more closely at the quantity of demand for each of their 17 devices, visualized below. Understanding the demand value of each medical center will be critical to accomplishing our goal of generating fair territories for the sales reps.

Demand Per Medical Device

Baseline Solution

Given that one of the company’s primary goals is to minimize rep drive time, we used a nearest neighbor analysis to build a baseline solution that assigns accounts to the rep who lives closest to each medical center.

While this solution is the most spatially “efficient”, given the uneven distribution of patient capacity and varied demand value previously discussed, we can expect that this solution may not meet the company’s second priority - evenly valued territories.

This expectation is held up when we dive into the data:

Baseline Results

Even though each of our territories is spatially efficient the accounts are not fairly balanced:

  • The total value of sales for each rep is not even.
  • The number of accounts are not even.
  • There is a huge travel discrepancy between reps.

Build More Balanced Territories

What we are facing is a Multi-Objective Optimization problem: we are trying to simultaneously optimize for multiple goals which, in this case, counter each other. The more even we make the sales territories, the more each sales person will have to travel, which means that we will never be able to find a solution that perfectly addresses all of the company’s conditions and goals. Our baseline solution was already optimized to reduce overall travel distance, so any solution designed to address the other goals by making territories more balanced in terms of value will be less spatially efficient.

Not to fear. We can still dramatically improve towars the medical device sales company’s goals with the help of spatial data science.

Solution 1: Minimum Cost Flow

One solution to create more equitable territories for our sales reps is to use a Minimum Cost Flow algorithm to define our territories. The algorithm is essentially trying optimize by finding the “cheapest”, or minimum cost, solution to a set of objectives. Here, our objective is to minimize the total distance traveled by our reps while also balancing the value of our territories. Using this algorithm requires us to preset an ‘array of demands,’ which in our case is the number of Medical Centers within each territory. To do this, we maintained the same number of Medical Centers in each territory from our baseline solution.

Using this algorithm results in territories with higher total sales values ceding more valuable Medical Centers to lower value territories, while continuing to minimize rep travel time. You can see this defined in the map below:

This solution is limited by requiring the quantity of Medical Centers per territory to be pre-defined, but it allows us to improve our territories from the baseline. With the Minimum Cost Flow algorithm, we are able to reduce the standard deviation of sales value across our territories by 17%.

While this method does an admirable job of increasing territory equity, there is a cost, this solution increases the total distance travelled, and the standard deviation of distance traveled, so while sales reps will benefit from more equitable sales opportunities, some will be travelling further, incurring more costs for the client company.

Finding a more equitable solution may require a bit more power.

Solution 2: Genetic Algorithms

Inspired by evolutionary patterns of mutation, crossover, and natural selection, genetic algorithms are an optimization tool that allow us to find the “best fit” solution to a particular problem.

This algorithm does not require the input of an ‘array of demands’ like the minimum cost flow method. This means that the results of the genetic algorithm are untethered to the medical center quantity per territory constraints set in the baseline solution.

Using this type multi-objective optimization, the output of the genetic algorithm is an infinite number of solutions that are all equally optimal - what’s called Pareto Optimality - and allows our medical device sales team to select a solution that fits their internal feasibility requirements.

Pareto Front
Pareto Optimality is a measure of efficiency wherein it is impossible to reallocate resources to improve one specific value or criterion without worsening another value or criterion. Our Pareto Front, as represented below, includes all possible solutions that match this description, thus providing an infinite set of options for an individual looking for optimization. That individual is left to choose which optimization best meets their situational needs.

The below visualization represents the territory solution created by the genetic algorithm that exists in the ‘sweet spot’ of the Pareto Front, where standard deviation of sales and distance travelled are most balanced.

While you may note that this solution has created some slightly larger territories, where, for example, a non Chicago-based sales rep may have a couple of high value accounts in the city, this solution is vastly more efficient and equitable than either the baseline or solution 1.

Using the genetic algorithm, we have reduced the standard deviation of sales by 50%, with only a 3% increase in total distance travelled, keeping travel costs largely in line with the baseline solution. In addition to creating significantly more equitable sales territories, the standard deviation of distance travelled is essentially unchanged from the baseline as well, so ‘fairness’ around travel time is maintained as well.

Keep in mind the algorithm is providing a near infinite number of solutions along the Pareto Optimality front, so if the client prefers to further reduce travel time or to level their territories by further reducing the standard deviation of sales value, they have the power to do so.

Now that the medical supply company has selected more optimal sales territories based on their preferences and defined parameters, they can continue to use spatial analytics to further optimize their processes. As a next step, they could also take their new territories and create optimized visit routes for each of their sales team, further reducing costs and maximizing rep efficiency.

Optimized Sales Routes

Reach New Heights With Data-driven Territory Management

As a sales manager, it’s easy to maintain the status quo with existing territories, often based on arbitrary geographies. Sales reps and the wider sales organization are likely comfortable, and disrupting that comfort may feel like more trouble than it’s worth. But a territory management strategy that isn’t data-driven is likely harboring unnecessary inefficiencies and costs.

By using data science methods, such as a genetic algorithm, to create optimized sales territories based on organizational parameters and goals, sales managers can greatly reduce these inefficiencies. These methods create tangible results, such as a 50% reduction in territory value inequity or significant reduction of travel costs, that can make a sales organization more equitable, more efficient, and more profitable.

NB: This example has been anonymized and data randomized for the purpose of this blog post.

Want to chat about optimizing your field sales?

Our experts are available to help you boost efficiency and achieve your sales management goals.

Request a Demo