Itemator

UX/UI, Mobile app, end-to-end design

Summary
Refining the process for receipt itemizations, processing and OCR in the Concur mobile app.

Receipt itemizations are one of the largest pain points for Concur power users.

Problem
Receipts, especially hotel bills are difficult to scan, which means users often have to manually input all individual expenses, from hotel lodging expenses to room service to small surcharges and fees.

The current user flow made it impossible to review itemizations before processing it to the cloud. The overall flow is slow, inaccurate and difficult to gather accurate data from. Most users, in their frustration, roll up all their expenses to save time.

Understanding the User
My research process included

  • Whiteboarding user journey maps and identified pain points
  • Interviewing users to describe their own experience through various steps
  • Collaborating with data scientists to understand their own issues with the current flow
  • Collecting user data on usage times, highest pain points & priorities

After research comes iteration
After fully mapping out the new flows and finishing user research I worked on,

  • Mapping the full user journey (Below)
  • Sketching new user flows and screen flows (Left)
  • Communicating with multiple teams and stakeholders throughout various iterations and steps of the project
  • Designing the final UI, based off the current Concur Mobile design library
I worked on this design from beginning to end, overseeing the entire process.

Outcome
The project started out as just hotel itemizations, but it eventually tackled an even broader issue: scanning & processing all receipts, more accurately and quickly.

By using more accurate OCR and continuously scanning multiple frames of a receipt, the confidence level* improved from 20~ to 80%.

Although the process still isn't perfect in detecting finer details such as vendor, date, city, it's a large improvement in terms of accuracy, speed and user experience.

Improvements include
  • Smoother user flow
  • Reduced processing time from ~10 minutes to under 30 seconds
  • OCR scanning confidence quadrupled
  • Easier to itemize, which means more usable data for user research and further improvements
Initial user testing came back with positive user feedback, users were especially happy with the faster processing times, as well as less time spent manually inputting their receipts. The data scientist team received more usable OCR data, thanks to more accurate scanning and input of expenses.


*Level of confidence reading the information on a receipt


Next steps
The app was for demoed for keynotes at Concur's annual conferences, including Fusion North America and Fusion EMEA. Further development has been made and the features has been put into production on the Concur beta mobile platform.

Go Back
Next