Pack App
WHY IS THIS IMPORTANT?
Amazon's brand promise is Earth’s biggest selection and being the Earth’s most customer-centric company. It follows that in shipping its products that it has become the gold standard for delivery you can count on. And as with any tech company, that success can always be improved.
During the holiday season, Amazon's warehouses employ contingent staff to support the volume of holiday demand. It can take almost to two weeks to ramp up new warehouse employees. As a new design team, the "Pack App" would be one of our high-impact projects for the year. We would redesign it to be piloted at one of the largest Amazon warehouses for the upcoming holiday season. Our initial goals were to cut costs with employee training and to reduce packing error rates.
THE TEAM & DELIVERABLES
Working with a design manager and senior designer, I focused on interaction design. We worked on a small interdisciplinary team with a copywriter and a developer. The senior designer and I worked together on the Pack App's design, conducted our own research and worked with the copywriter and developer to create prototypes and iterate.
While I worked with a team, I personally created sketches and visuals in Keynote to pass on directly to anyone on the team. This included type hierachy, color and field testing these design details. As a team we used Keynote as everyone could contribute to the document and it could easily be presented.
We worked quickly and within 2-week sprints. At the end of the sprints, design would draft a testing and moderation guide and visit a warehouse for testing. There were several Amazon warehouses with whom we had close relationships and were accommodating to this two week cadence.
Millions of books, and hours of testing logged
Getting our developer involved in testing on our trip to the Reno warehouse
CHALLENGES & SOLUTIONS
When I joined the team, the Pack App software was in a good starting state. My senior partner had interviewed developers and subject matter experts. But it still needed to be designed and tested end-to-end. And about halfway through the project the senior designer would leave for another opportunity. I was able to see the design and testing through launch with the support of my design manager.
Originally, the software users were packing orders from command line, a basic navigable interface. Having a GUI was a dream. But they were still having issues with the visual interface. Why? Coming into the project with a brand new set of eyes, I was ready to dig in.
Besides ramping up new staff more quickly and reducing error rates, design also wanted to make the code more reusable and consistent. We were moving away from a command line approach but the software was still meant to be function over form.
The original design
Left to right
Users read left to right. The original designs (Top right) weren't leading the eye. Was there a way we could make sense of the content? We showed users where they were in the process with breadcrumbs and grouped similar content. Such as, the items to scan go into box "J5."
Users aren't staring at the screen the whole time
Part of the packing process includes looking at the screen, seeing a list of items that needs to be taken from a plastic bin, or "tote." Checking that the number of items on the screen is the number of items in the tote. Checking that each item in the tote is in fact what's on the list on the screen. Putting those items into a box of a particular size. Here, it's size "J5".
That small slice of actions alone means the user isn't looking at the screen the entire time. Losing your spot on the screen means that it needs to be clear that certain areas can be counted on for specific information. Does bigger mean better? How helpful is highlighting relevant information? What type of highlighting should be available? What about movement or animation? We tested these hypotheses to see which worked best in which situations.
Ultimately, color was used sparingly. Color is used by warehouses to help users find boxes quickly. Each size box has a color associated with it, found below the packing table. Highlighting items is a great way to show what to scan next. As well as allowing the user to move through a list with their arrow keys. And scanning the label at the end of the entire process, blinks after scanning the last item in the shipment, Don't forget this!
Audio cues
Similarly, when a user gets really good at packing, muscle memory takes over. Even really good packers with low error rates can pack at such a high speed that they need only glance at the screen to get the information they need. In those situations, audio cues are great. Used successfully, they can signal task completion. Another consideration is competition from general warehouse noise. Testing the audio punctuation and volume was also in our testing guide. Audio cues were used to signal initial scanning of the tote (unique), after each item (unique set) and to complete the tote (unique).
After a user has scanned a "tote," or customer order containing 21 items
The packing process for this "tote" is complete
What's Next?
The Pack App was implemented in the Indianapolis Amazon warehouse during the holiday season to success. The ramp up time went from 2 weeks to a matter of minutes with negligible error rates. The most exciting part of testing was when we would ask employees from other warehouse processes to pack for us. Without any traning or context and being led entirely by the UI. Following this project, two other packing processes were scheduled to be reviewed for update by our team. Including a version of the app we designed, to be used on scanning devices (scanner and wearables) for other packing use cases.