Project Yellow.
How might we make access to clean, structured data easier while honoring Excel’s promise of programmable data in a flexible grid?
Problem
Project Yellow is a multi-year effort to expand Excel capabilities as a programming environment. For all the power of Excel, there are still fundamental concepts missing. Over the past 30 years, there have been many innovations in the grid: sort/filter, conditional formatting, pivot tables, just to name a few. [But] the content of a cell itself is still no more powerful than what you can put into notepad.
A key piece of Project Yellow addresses this with new data types. New record data types enable a single cell to contain multiple data, expanding the programming capabilities of Excel.
We also had a $4M / yr deal with Wolfram Alpha that our leadership team wanted to see results for. The product team was very siloed, shipping individual features that didn’t flow together seamlessly. We were also designing for Excel, Microsoft’s most loved product, but a complex legacy software system full of technical constraints that make change and innovation very challenging.
Our goal was to ship Wolfram Alpha data types to General Availability and to increase our funnel numbers. The silos on the team to date had led to a user experience that was unintuitive and fragmented.
Actions
I joined the Project Yellow team as the lead designer in 2019, directing the work of three junior designers and coordinating the product backlog with three PM and two Engineering leaders. We had an existing MVP but our usage numbers weren’t great: we had a huge drop-off in the first stage of the funnel.
Here are the actions I took:
I assigned different areas of work to the younger designers based on their skills and interests.
I made customer needs the focus of our product planning, pulling existing research together, prioritizing Jobs To Be Done around this work, integrating them into the new weekly product team critiques.
I developed design principles and strategy that had not previously existed and showed the team how to perform heuristic analyses to identify gaps in the user experience.
I worked closely with my Principal-level PM and Engineering partners to transition the team into working in a true multi-disciplinary model.
I spearheaded Design collaboration with teams across Office to consolidate design work and ensure a consistent user experience. Specifically with our SKI team (Search Knowledge Interface).
I then took the usability gaps identified via our heuristic analyses and led the designers through a vision exercise, building end-to-end north star prototypes exploring solutions to our UX gaps. These prototypes were used to inform our current product backlog. I can only show small parts of this work due to NDAs, but here are some of the heuristic violations we found.
Results
The minor usability improvements we made to the UI based on the heuristics evaluations I led resulted in the highest customer satisfaction ratings since the initial MVP release which happened a year before I joined the team. We did move some of our metrics, but we now know based on quant + qual data that for our records data types to be widely adopted, data scope and quality is paramount. People also want to build their own data types by connecting to their own data sources. So we pivoted our strategy and modified our focus to address user pain points surfaced in our MVP diary studies. The Project Yellow work continues, but Microsoft ultimately made the decision not to renew the Wolfram Alpha contract.
Press
Inserting Images in Excel
Excel Data Types: stocks and geography
Jobs To Be Done Framework
Know Your Customers Jobs To Be Done