Mobile Receipt Capture
In spring of 2015, I worked on a project to rethink Concur's model of capturing an expense on the mobile phone. There were three main problems we were trying to solve for: addressing that users often did not think to open the Concur app to take a receipt picture (because it was easier to just take a photo with their phone's native camera app), grouping images together, and the overwhelming task of creating an expense from a receipt.
We needed to figure out at what point in the process it made sense to ask if the user was going to upload multiple images and group them as part of one expense (as is often the case with long receipts when the entire receipt will not fit--and be readable--in one photo). After illustrating the impact of various options, we decided to handle image grouping after the user had selected their photo(s) to upload. This also allowed us to address the issue of users opting to use their native camera app on their phone, because this order of tasks made it easy to include the ability for users to choose photos from their camera roll, similar to Instagram.
In order to help with the overwhelming and confusing expense-creation process, we decided to decouple the concept of receipt capture from creating an expense, while also using the receipt image as the starting point for starting the expense process. I had the idea to treat the receipt photo as the "front" of the image, and create a digital "back" that contained the expense details, as though the user was recording those notes on the receipt. This allowed users to toggle their view easily, to reference the receipt information while filling out the expense (a common request from our users.) Because we did not have OCR capabilities at the time, we sought other ways to ease the burden of data entry for our users, which was a significant prohibitor for adoption of using the mobile app for creating expense. Some of the features we added to help with data entry included using the appropriate keyboards for each field (text for alphanumeric fields, and number keyboards for fields which contain only numbers), and incorporating some predictive text capabilities on form fields.