Home Depot wants to improve their FirstPhones so in-store employees can work faster and happier, and more efficiently help customers. I put together a user research team to investigate users and stakeholders, went through design iterations, and arrived at a final prototype. 4-month long class project at Georgia Tech with input from Home Depot corporate.

FirstPhone basics

A FirstPhone is basically an Android phone running custom OS. Its case also serves as a barcode scanner. Almost every Home Depot store employee uses the FirstPhone, though for vastly different purposes. A regular associate may use it to check inventory information, while a manager can use it to check sales figures. It is also worth noting that FirstPhones are shared between employees, since there often aren’t enough of them in a store.

The FirstPhone currently offers 40–45 apps, though many employees often use only a small number of them in their daily work.

The FirstPhone’s home screen, captured during one of our store visits.

Investigation

We conducted desk research, field observations, contextual inquiries, semi-structured interviews, and interim user testing to understand the issue at hand and come up with design directions. We also used survey results from Home Depot’s subreddit to complement our qualitative work.

Problems and personas

During our investigation we gradually learned the way Home Depot develops and maintains FirstPhones and its corollary issues. Since FirstPhones run on multiple legacy backends and must be online almost all the time, modifications are often delivered as hotfixes or new apps in a one-off fashion.

This way of working results in three big problems. First, 40 apps are way too many (logging out should not have to be its own app!). Many of their features also overlap, meaning there are often multiple ways to do one thing. Conversely, features involved in one task can also be spread across multiple apps, requiring the employee to jump from one to another. These problems make working cumbersome even for veterans, let alone the day-one employees in stores with high turn-over rates.

The obligatory staring-at-post-it-notes pic.

We derived three personas from our articulation: day-one associates, veteran associates, and associates in managerial positions. While day-one and veteran associates differ in their experience, tech-savviness, and potentially socioeconomic status, they share the common objective of completing their tasks well. Managers on the other hand are more concerned with keeping the department in good hands and keeping their people in good direction.

Design directions

Our first direction sought to alleviate the jumping problem by deep-linking the user into all the apps he/she needs to complete one task—effectively a task-based app launcher. Since it would be just an Android launcher, this direction also happens to be the most convenient to realistically implement.

Another teammate made an Adobe XD prototype for later testing. Tapping see all apps always reveals the existing 3×3 app grid.

Inspired by iOS’s Siri Shortcuts and the Workflow app, the second direction systematically rethinks the relationship between tasks and employee workflows. All apps would be broken down to their constituent functional modules, such as “scan an item” or “make a PSA”. They are then aggregated in a library, ready to be picked and recombined to form any workflows imaginable. This direction most competently addresses employees’ different customary working styles, which have emerged partly as workarounds to the FirstPhone’s cumbersome experience. Once fully implemented, it would be the cheapest to update and maintain, since all changes can be expressed in new modules added to the library.

Upon signing in, the associate is greeted by workflows, not apps. Some workflows are pre-installed by Home Depot corporate or in-store managers, while others can be fully customized to his/her liking.

I made an Origami prototype to feel the drag-and-drop interaction.

Our third direction is a new take on FirstPhone’s existing scanners. Based on augmented reality and object recognition, employees can immediately see the most relevant controls and features associated with items, from which are what a majority of the tasks originated. AR is also useful in providing additional guidance in non-item-related tasks such as in-store wayfinding, indicating a potential integration into Home Depot’s consumer app.

Concept sketches made in Sketch and Illustrator to demonstrate our idea.

After we made low-resolution prototypes, we arranged a round of user testing with a Home Depot store in Atlanta. The second direction was quickly dropped as employees had significant doubts as to whether they are tech-savvy enough to utilize it to their full benefit. We eventually narrowed down to a final design: augmented-reality-assisted task management.

Final testing

Towards the end of the semester, we were invited to Home Depot’s Atlanta headquarters to do heuristic evaluation and user testing with their designers and other stakeholders. Since we spent most of our semester doing qualitative field research, the end prototype did not have the formal polish it deserved. Our testing results confirmed this—there are still significant issues with basic layout, information architecture, and even iconography.

My inquisitive teammates gazing at our test subject at Home Depot headquarters.

Closing

Like most app design projects, our prototype still needs to be refined. But if there is one thing that stayed with me for the longest, it would be the realization that design cannot be solely considered in terms of the end artifact; instead, it needs to be considered around people—not just people who eventually use it, but also people who mediated it and made it happen. The way FirstPhone looks and works is a testament to how people maintain, develop, and ultimately see it. As designers, we then need to focus on not only those who use it at the end, but also those who are involved in the process of making. Only with such focus can we truly inject human-centered values into our work.