I did
Redesign of the Philips Xelsis automatic coffee machine, with an emphasis on user experience
Philips sponsored project
@ TuDelft
UX Research
UI Design
Interaction Design
What We Made
We delivered a full redesign of the Philips Saeco Xelsis fully automatic coffee machine.  We were able to drastically reduced errors, provided a more personal brewing experience, and cut time to brew a coffe in a third. This project was done during my exchange to the master Interaction design program at TuDelft.
Usage Inspection
Design Process
Task Flow Mapping
Market Research
A-B User Testing
Customer and Stakeholder Interviews
Coffee Drinking

Needs a quick cup of coffee, without having to worry about consistancy.

Quality and prestige

User is a coffee afficianado, but also cares about prestige of the brands he buys.

Core Customer Needs

Needs a variety of coffee options, mainly for his guests.

Philips coffee machine redesign
Usage Inspection
Tasks such as creating a profile, or brewing a cappuccino were broken down into individual touch points. Pain points within these tasks were flagged.
Finding some quirks
We started with a thorough usage inspection of the machine. We analyzed its individual components, the machine’s functionalities, as well as general maintenance procedures.  This allowed us to become very familiar with the machine, as well as many of the nuances and quirks associated with using it on a day to day basis. We gleaned insights to problems regarding the user interface, the fullness indication of the drip-tray and locating the different compartments. Therefore, these are the areas within the machine that we decided to focus on throughout this project. 
Usage Inspection
Prototyping in 3 directions
We diverged our preliminary ideas into three categories: Safe, Bold, and Unexpected.  Safe ideas addressed problems, adhered to all constraints, and were somewhat expected in nature because they may have been done before on similar machines.  Bold Ideas addressed the ideas, but pushed the constraints a bit in favor of making something more original.  The ideas we had that were fairly abstract in nature and might not respect all constraints were labeled as Bold. In respect to these three directions, testable prototypes were built for drip tray variations, compartment variations, milk systems, symbols, and digital interfaces.
Diversifying Ideation
A/B User Testing
Round 1
Initial user testing of our three concept directions. After going through a series of tasks with each machine, images were used during the final interview phase to help as visual aids for the participants. By leading participants to a separate space in the testing room away from the prototype, we allowed them time to reflect on the portion of the test that was just completed while a member of our team reset the machine for the next task out of their field of view.
Testing our directions
Desired daily coffee
We found that participants often prefered the same coffee each day, just the way they liked it.  Our concept of a mug linked to their favorite coffee made the brewing process much quicker, simpler, and more enjoyable.
Providing a shortcut
Based on our insight into daily drinking habits, we designed an RFID based sticker that goes on the bottom of a favorite coffee mug and provides a shortcut to the user's favorite coffee. This fulfilled the user's needs of a quick cup of coffee just the way they like it, without sacrificing other coffee option for guests.
Too many choices?
KJ Method
We analyzed the results of the tests using the KJ method.  This meant coding the concepts A, B, and C to three colors of post-its, and noting anything that stood out during the test.  These notes were organized in categories they related to, such as the drip-tray, compartments, and UI.  By analyzing trends between multiple users, we were able to select the parts of our concepts that users found worked the best.  These successful design elements we then used as the basis for the next iteration of our prototype for further testing.
Milk carton shortcut
By observing the way people used the machine, we learned that many users would ignore the milk carafe, and simply use the milk carton instead.  This saved time and milk.  This insight made it’s way into our final redesign.
Our redesigned milk carafe
Piecing together what worked
Based on our findings, we combined the most successful elements of our 3 concepts into a new testable prototype. We found users liked the pedestal-style drip tray for understanding the waste water level, although seeing the water was a bit ugly. The compartment windows on the water window and coffee grounds were nice to insure enough water was present before starting the brewing process, as well as understanding what was in each drawer.  Participants liked the integrated milk system that attached the carton to the side of the machine, as it looked better and had fewer components to clean.  An interface which linked to mugs for a quick and personal coffee in the morning was also desirable.  We developed all these elements a little further for another round of user testing.
The Next Iteration
"Can you please make a small cappuccino?
We recruited a new set of participants for a final round of user testing.  They all closely matched our target persona “Bernd,” meaning they were between the ages of 35-55, working professionals, and made coffee 3+ times per week for themselves and guests.  These participants were given a range of tasks relating to various elements of the machine we had refined, with the aim to test their usability against the previous versions.
User Testing
Color coding participants
All participants were recorded through video and audio.  Once we had concluded testing our new prototype, it was time to organize this data.  Similar to the first round of testing, we used post-Its to record notable errors as well as positive comments.  We used a variation of the KJ Method this time, coding each participant to a different color Post-It.  These were sorts based on the area of the machine they related to, and similar notes were stacked.  This made it very easy to see trends among all the users.
Analyzing Results
Water as the indicator
One of the main complaints of the original drip tray was that users couldn't tell when it was full, leading to spills and frustration. We redesigned the tray so that excess water would flow around the tray in a mote, making it much more apparent when it was time to empty.
A redesigned drip tray
Constraints of a redesign

Being a redesign of an existing product, there were inherantly many constraints and challenges.  It was very necessary to fully understand the market segment, the users for this specific machine, and manufacturing limitations.

Benifits of thorough documentation

Being a school with a strong engineering background, TuDelft taught me the benifits of well documented research and design process.  Through detailed reporting of our findings, we were able to navigate our work efficiently to produce the best results.

Key Learnings
Working with the Dutch

Working within a diverse team for a Dutch client was a great lesson in multidiciplinary teamwork and cross cultural collaboration.  It quickly became very apparent that I needed to adapt my methods of communicating to understand the “Dutch Direct” way.

The team at Philips HQ in Dracheten: Yu Liu, Hannah Pak, Henrique Meyrelles, Me, Lisa Kleisen
Decreased brewing errors, increased brewing satisfaction
After presenting our redesigned Xelsis to Philips at TuDelft, we were invited to their headquarters in Drachten (Northern Netherlands) to share our work with more members of the company.  We were met with very positive feedback, and were asked to share our work with their Milan branch as well.  It was a very positive experience working with Philips, as well as the diverse and international team at TuDelft.