Robert Boler
service + software designer

Investigating Lost Shipments


Investigating & resolving customer orders at a glance

Unifying features from seven disparate systems to create a better employee experience, cut resolution times in half, and launch a predictive triage algorithm

  • Client a global logistics company

  • Roles interaction designer, user researcher, presenter

  • Duration 3 months

  • Traits #artificial intelligence #employee experience #product design #web app

Note: Details of this case study removed for client privacy. Contact me for more on this and other projects.

“Where is my order?”

49 times out of 50, our client’s customers don’t have to ask this question. But sometimes one of their tens of thousands of orders gets lost or misdirected, leading to customer frustration and an average four-day investigation and resolution period. Our mission: to cut down resolution time, improve their employee experience, and become more proactive in addressing problems with orders.


The existing system had agents logging into and switching between seven disparate systems.


Finding the sweet spot between “too little” and “too much” information

  • Conducted over a dozen on-site user interviews and usability test of a prior concept, taking note of unspoken insights and their physical workspaces

  • Laid out the seven legacy systems they currently use side-by-side to compare data and construct the story of a problematic order

  • Observed how they communicate with their customers and asked how they wish they could

  • Plotted the many different types of data (timestamps, location data, images, facility information, order details, customer details, and more) along the axis that best helped the investigative process: time

  • Shaved off irrelevant or cryptic order information, while making details accessible with a single click

  • Worked with our developers to construct a predictive algorithm based on historical data, and designed the presentation of that data to point the agent in a helpful direction without being intimidating or implying false accuracy


Our new tool grouped related information, created a cohesive timeline of the order + support ticket, and used past resolution data to point the agent in the right direction without decreasing autonomy.


Challenges converting raw data into actions

  • Surfacing the right information in the best contexts

  • Communicating algorithmic insights to the agent without removing agency or implying false accuracy

  • Helping our main stakeholder build client-side support for the product against bureaucracy and complacency


We measured the helpfulness of the triage prediction with a user survey, displayed at case resolution.


A domino-effect of better experiences

We were able to combine seven different antiquated resources into one focused, flexible tool that presented information contextually, proactively provided causation insights with an ~80% successful algorithm, dramatically improved the employee experience, and cut resolution times from four days to less than two.

80 pct@2x.png

This project directly led to at least four more streams of work with the client, including a 2.0 version of the tool and a downscaled version used by the entire company.

(Header image by Ivan Bandura.)


To learn more about this project, other design work, or random bicycle trivia,

/  holler at me.