Americans’ demand for delivery services has surged amid the COVID-19 pandemic. With the widespread adoption of social distancing, the need for reliable and efficient delivery of essential goods continues to climb. Pharmacy delivery services, which deliver prescription medication to patients’ homes, are especially critical because they serve those patient populations that are most at-risk and most impacted by social distancing.
ScriptDrop is a prescription delivery service that uses a network of couriers to deliver medication from pharmacies to patients’ homes. As demand for prescription delivery skyrocketed due to the COVID-19 pandemic (see Figure 2), ScriptDrop expanded their services across the country and condensed their long-term growth plan into just a few weeks.
Every courier in ScriptDrop’s network is assigned a set of pharmacies and is responsible for all deliveries originating from those pharmacies. Previously, ScriptDrop relied on a manual process for pharmacy assignment that could not scale with the soaring demand for their services. ScriptDrop partnered with Mobikit to create a fully automated, data-driven pharmacy assignment model that allows them to scale rapidly while meeting business objectives and operational demands.
Why Pharmacy Assignment at Scale is Difficult
An effective method for pharmacy assignment at scale must be able to match tens of thousands of pharmacies to hundreds of couriers while taking several business-critical requirements into account:
- Respect courier constraints around geography and capacity.
- Meet projected demand across pharmacies and patients.
- Serve any potential customer regardless of location.
- Improve the customer experience by rewarding high performing couriers with more desirable pharmacy assignments.
In this section we’ll describe some characteristics of couriers and pharmacies that present significant challenges when developing an assignment model that achieves these goals.
No Two Couriers are Alike
Different couriers have different service capacity and may have a variable number of drivers available each day. Over-assigning a courier on any given day may cause delivery delays. Courier quality is variable; high quality couriers are preferred because they promote patient satisfaction and reduce stress on ScriptDrop’s customer call center. However, using lower quality couriers is sometimes necessary to ensure adequate coverage.
Pharmacies and the Urban-Rural Divide
There are nearly 70,000 pharmacies in America and they vary greatly in the number and geographic distribution of patients they serve. Delivery services are easier to set up in areas with high population density, but ScriptDrop wanted to ensure that they can offer their service to patients everywhere. The biggest challenge to achieving this goal is the urban-rural divide.
Let’s use Ohio as an example. Figure 3 is a map of Ohio pharmacies overlaid on census population data. There are nearly 3,000 pharmacies serving nearly 12 million patients across the state. Even though 44% of Ohioans live in rural areas with low population density, only 31% of Ohio pharmacies are in these areas (see Figure 4).
This imbalance is exacerbated by the geographic distribution of rural patient populations. In other words, not only do rural pharmacies serve a disproportionate number of patients, but those patients are distributed over a larger area. In urban areas, approximately 90% of people are within 2 miles of a pharmacy. In rural areas, however, this threshold jumps to 13 miles (see Figure 5). As a result, delivering from a rural pharmacy requires significantly longer travel times, which makes rural pharmacies more expensive for a courier to service than urban ones.
This is a critical consideration when structuring incentives for couriers. If a courier is assigned too many rural pharmacies, it could impact their profitability. On the other hand, couriers must be incentivized to accept some rural assignments so that ScriptDrop can offer its services to rural patient populations.
Historically, expanding ScriptDrop’s services into new markets typically required months of human effort to analyze the data on local pharmacies and couriers and manually generate pharmacy assignments. However, as demand for ScriptDrop’s service grew rapidly each day, this approach quickly became unsustainable.
How Mobikit & ScriptDrop Collaborated on a Solution
At the onset of the COVID-19 pandemic, Mobikit partnered with ScriptDrop on their mission to expand their critical services across the nation. Mobikit helped Scriptdrop develop a fully automated, data-driven pharmacy assignment model that:
- increased patient satisfaction and reduced call volume by prioritizing higher quality couriers;
- provided equitable and explainable pharmacy assignments to couriers while accounting for courier capacities, geographic constraints, and balancing the allocation of urban and rural pharmacies;
- updates in near-real-time to account for rapid surges in demand and fluctuations in driver availability.
While multi-objective optimization models like these are not new in the field of operations and logistics (, , , ), there are underlying technical challenges that make their implementation difficult.
- A model is only as good as the data that drives it. Hence, the lion’s share of effort in solving problems like this is typically spent on data engineering: cleaning and standardizing the underlying location data and enriching it with meaningful features.
- From a business perspective, any model used to drive business decisions cannot be a black box: it needs to be explainable to couriers, pharmacies, and internal stakeholders.
- In the long term, stakeholders need to be equipped with proper tools to monitor and improve the model.
Through Mobikit’s interactive web-based platform, ScriptDrop’s logistics team was able to rapidly develop and validate a viable and explainable courier delivery model, without having to worry about the underlying data engineering challenges. Mobikit’s location-focused data preparation tools automated the journey from raw data to actionable insights and a working model. Mobikit’s big-data visualization features allowed the ScriptDrop team to inspect and validate the model’s output before deploying it.
Delivery services are playing an increasingly critical role in our world. As our partnership with ScriptDrop demonstrates, Mobikit’s tools have tremendous potential to optimize delivery operations across pharmacies, grocery stores, restaurants, and other essential businesses. By eliminating the technical barriers to automated, data-driven decision making, Mobikit can empower delivery services to scale rapidly and effectively.
Interested in learning more? Reach out to us at email@example.com.