Grubhub | Courier Scheduling

Offering delivery services to restaurants all across the country is no short order. With hungry customers at all hours of the day, it takes a whole team of people to ensure markets are appropriately staffed to meet expected demand.

At Grubhub, I designed a new internal scheduling tool that can scale with the company as delivery services continue to expand. The result is a solution that enables the delivery operations team to quickly staff the right number of drivers, evaluate market health, communicate with drivers, and track historical data. On top of improved work efficiency, Grubhub was able to save over $130k annually in vendor costs.

Grubhub partnered with When I Work (WIW) to offer a scheduling solution for the independently contracted couriers. 

In a nutshell, WIW is a platform designed for small businesses like restaurants or independent retailers. It offers a flexible way to provide open shifts (or blocks, as Grubhub calls them) that staff can sign up for.

This was an ideal solution for a company starting a delivery business, but as Grubhub grew, it became obvious that it was time to create a more custom, efficient solution.

The image below is a screenshot from When I Work. When looking at a single day, the product lists every single driver in a market, regardless if they’re working that day or not . In many markets, there are thousands of drivers, thus resulting in endless scrolling and a time-consuming scheduling process.

Creating a robust custom scheduling tool isn’t easy. In order to grasp the complexity of the project, I needed to thoroughly understand the current day-to-day workflows and pain points of the internal users.

I scheduled both interviews and shadowing sessions to observe how individuals in various roles at Grubhub used the existing scheduling tool. The primary users, Driver Specialists, were in charge of setting up availability for every delivery market each week, and to make sure there is enough coverage for each day.

I took my findings and shared them with stakeholders from operations, product, and tech. I recommended that the pain points outlined below were problems our new solution needed to solve.

• Total coverage on an hourly basis was more important than knowing which individual drivers were scheduled
• WIW is slow, clunky, and taxing on a computer’s resources
• With thousands of drivers, the scrolling felt nearly endless
• Cannot assess market health at-a-glance
• Scheduling tools aren’t flexible enough to support unique use cases with driver scheduling
• Hard to keep track of driver actions (picking/dropping/missing blocks)

After interviews, shadowing, and a thorough understanding of the existing tool, I was comfortable outlining the information architecture (IA) required for a successful scheduling tool. I validated this with the project stakeholders, and went to work on the fundamental components of a schedule.

A schedule represents the sum of its events, or “blocks” as we called them. The block component represents a period of time in which a courier is on duty for delivery. A single block can be offered to different vehicle types (e.g. car or bicyclist). A block is almost always available in multiple quantities (e.g. needing 50 drivers on the road between 11am and 1pm). Driver Specialists tend to spring into action when there aren’t enough drivers signed up for a block, so this data point also needs to be represented.

A single block representation needs to be data dense, flexible, and easily repeatable. After multiple iterations, I came up with the design of a single block below. Next to it are variations of a block, showing its flexibility by representing different data.

The scheduling tool most closely resembles a calendar, utilizing both day and week views with ease of navigation between date ranges. It also behaves like a calendar in that subsequent actions usually occur within context of the main content. This is why most of the courier scheduling actions appear in the form of a sidebar, which appears to the right of the schedule (as seen in the image below).

Driver Specialists need to be able to manipulate blocks, view or message couriers, and review actions taken on a block.

Thanks to the relative ease of access to our intended users, the entire design process was iterative with a healthy feedback loop. With each new milestone, I ran down to put the design in front of Driver Specialists in order to capture feedback.

After a solid set of features were completed, we invited several Driver Specialists to user test our InVision prototype.

Our participants were enthusiastic about the speed and simplicity this new scheduling tool offered them. We learned some valuable information and gained even more insight during these observations. For example, Driver Specialists indicated that when blocks are full, they’d like the ability to filter them away so they can focus on the problematic ones.

The image below shows an example of the day view. On this image, there is a breakdown of total number of couriers by vehicle type at the bottom. Each vehicle type matches to a color for easy scanning. Additionally, the requested filter to hide filled blocks is in the toolbar.

These custom scheduling tools not only increased satisfaction among its daily users, but it also greatly increased work efficiency for them. The gain in efficiency amounted to user cost savings, and major expense savings by replacing a vendor that was charging over $130,000 annually for their services.

Why stop now? Check out more of my work