Logo design by Ryan Koziel

The Power of Insights: A behind-the-scenes look at the new insights platform at Uber

Etienne Fang
Published in
10 min readAug 27, 2019


In the late 20th century, the dominant technological question was “How do we learn better?” In the early 21st, that question has been followed by “How do we better use the things we’ve learned?” Uber is powered by information. Chances are good that your company is, too. We’ve all had firsthand experiences of an inconvenient truth: information is only as good as our ability to make use of it. If a roadmap is locked in the glove compartment, no matter how detailed it is, it’s not much use on a road trip. Managing, sharing, and learning from that information is the key to using it successfully.

Around the world, our engineers, researchers, designers, product managers and every member of every team learn new things that can make Uber better. We want to be sure that these insights could be easily submitted and shared so that a team didn’t waste time relearning in Mumbai what another team in Mexico City has already experienced.

That’s why we created Kaleidoscope.

This challenge is by no means unique to Uber. At your company:

  • Do you want to make an impact with your insights?
  • Do you need to find existing information and domain experts?
  • Do you want to share insights from a specific project more broadly than with your immediate team?

Learnings, data, and ideas — when put together — represent dynamic patterns, as varied and colorful as the view through a kaleidoscope. Where a telescope offers a single long view and a microscope an in-depth, narrow view, a kaleidoscope triangulates changing views to create new patterns. Like a kaleidoscope our new insights platform allows teams to come together to create a macro-view that takes into account varying sources and perspectives. It’s through seeing and making sense from these changing patterns that we can perceive new images and pursue new possibilities.

Kaleidoscope UI design by Charlie Waite and Stephanie Brisendine.

Sharing vs. silos

At Uber, we are working to embrace insights across all levels and functions and strive to learn globally and diversify our perspectives in order to develop an effective long-term strategy.

At some companies, the science-based disciplines of research — including user experience research, marketing research, and data science — are viewed as more strategic, while operations-based insights from customer support and regional operations are seen as more tactical. They are frequently seen as separate categories, with little cooperation required or even desired between the two. Fortunately, we don’t feel that way at Uber.

I lead the Insights Platform, a group within User Experience Research. Our mission is to increase the impact of insights globally to inspire and inform Uber’s vision and priorities. The foundation of the Insights Platform is the belief that insights from all functions globally are valuable and that methodologically sound and valid insights can come from many places. In other words, we all have a lot we can learn from each other.

From our team’s inception, we knew that organizations across Uber would be “smarter together”, rather than remaining in “knowledge silos.” Our aims were simple:

  • Share insights from a variety of teams around the world
  • Inspire and inform the company’s decisions on where to invest resources

Simple, maybe…but not easy. Here’s how we set about accomplishing our goals.

Our Process

These are the steps we took to get to where we are today.

First: Developing our strategy

Our audacious goal was to build a holistic insights solution that could consistently deliver global and actionable insights to inform Uber’s priorities. While the expectation from stakeholders and leadership was to create a database to house insights, our strong belief has been that without mobilizing the necessary people and methodologies to gather, synthesize, and leverage those insights, the database would be ineffective. You can build the biggest library in the world, but without librarians and a card catalog, it’s pretty useless: without the people and process to leverage the database, impact would fall short.

Components of Uber’s insights taxonomy.

Second: Establishing a unified insights taxonomy

Our first step was to develop an insight structure that anyone could use to import or export data. Because insights come from a wide range of data and sources, with functions responsible for different types of insights, we needed to define a unified way to navigate the data.

After reviewing more than 800 insights previously gathered in one planning cycle across the company, we noticed a few problems in the way insights were considered and presented. Insights varied in level of granularity and ran the gamut from a few words to a paragraph in length. Facts were often presented as insights without rationale, context, or explanation. A lot of insights were tactical recommendations to fix bugs, which, while important, did not address the larger issues and ignored the potential for a deeper, more lasting impact.

From this analysis, we established the insights taxonomy for Uber, striving for simplicity, consistency, and actionability. The insights taxonomy consists of the following components:

  1. Insight statement: A succinct expression of what has been learned based on facts. It should be written with the syntax of “who + where + what + why” in order to be descriptive enough to stand alone and invite the reader to learn more.
  • The ‘what’ explains the behavior, occurrence or situation that is notable or problematic.
  • The ‘why’ explains the assumed reason for a behavior, occurrence or situation. From our audit of previous insights, the ‘why’ was most commonly missing from insights, and also the most critical part. The ‘why’ is the learning or takeaway.

2. Facts: Data from surveys and experiments that support an insight; anecdotes, user quotes and observations from qualitative research, analysis from customer support.

3. Opportunities: Recommended actions based on the insight. These can range from tactical to high-level “how might we” statements.

Sample Insight Statement before applying insight taxonomy vs. insight after applying insight taxonomy. (Facts and Opportunities not shown).

Third: Understanding internal users’ needs and motivations

The insights database exists to help Uber employees build better products and programs. We sought to establish a foundational understanding of our users through work sessions and user research, eventually arriving at two key user groups to design for: “insights creators” (who generate insights from various sources) and “insights consumers” (who use insights for learning and informing decisions).

We conducted several rounds of initial 1-on-1 interviews to understand the common needs and motivations for each group, and then followed up with iterative usability research. Users continue to help us prioritize feature development for the tool and our program.

Fourth: Identifying partners, piloting processes

Since this is our first cross-functional tool for insights and document discovery, we needed to establish an ongoing process for engagement. We conducted a few pilot programs to gather insights with our team of 90+ UX researchers. The purpose of these pilots was to test out the insights taxonomy, the approach without any custom tools, and the type of guidance users needed. Once we defined the process of submitting insights, we established a network of insights creators who shared common beliefs about knowledge sharing.

Fifth: Developing the minimal viable product, then iterating

Concurrent to the pilots, we built a simple MVP using external database software. With our MVP, we were able to bring to life the concept of the tool. Through some initial trials, we realized that the software itself posed too many usability limitations.

Through iterative user testing, we developed a design brief for the next iteration, which dramatically improved the user experience of our insights tool, generating awareness and excitement from users across the company at a company-wide launch. We continue to develop important features and functionality guided by our original vision and based on feedback from users. We are also integrating our insights database with other internal collaboration tools to optimize engagement and grow our platform of tools to serve existing workflows.

Video designed by Brian Waddington

Principles of Uber’s Insights Platform

As our team moves into the second half of the year since launch, we reflected upon the learnings we have gathered from our successes and challenges so far.

Connect systems, rather than supersede

Each team and subteam at Uber has its own system of organizing insights and documents, as well as their own ways of sharing out their work to stakeholders. Some systems were already quite advanced at our launch, while others were just being created.

Many groups asked if they should override their existing system with our insights database. Others were reluctant to use another system, as their own was sufficient for their team’s purposes and challenging enough to maintain.

I use the mental model of “connecting strings of Christmas lights.” Each team is already shining bright in their own worlds. The most value we add is to connect these nodes and create a network that intersects. Never underestimate the power of connection.

Context and interaction are key to collaboration

While our focus has been to create an efficient, self-serve tool for sharing and discovering insights with supporting documentation and authors, we learned quickly that what users — researchers and product managers alike — desired was real-world, face-to-face connection with others exploring the same topics.

In early user studies for the insights database, users described wanting to find other subject matter experts to talk to. These experts could be from other functions and other regions who have knowledge and experience on their topic of interest from the past or present.

We learned that having context around insights and data gave contributors a sense of trust that information would not be misinterpreted. Context comes in the form of having robust supporting documentation, experts to follow-up with, and the assurance that the information is vetted for quality by our insights analyst who reviews each and every insight for consistency.

Triangulating insights drives focus and perspective
Deep knowledge on a particular user type, domain or geographical region through a single lens is highly valuable and actionable. Being able to aggregate and synthesize that deep knowledge from across functions and regions provides a holistic perspective on problems and opportunities that are more broadly addressable.

Validation and visible outcomes help sustain engagement

Building new organizational habits is difficult, particularly in a fast-paced, globally distributed company like Uber. Asking research, data science, and operations teams to share insights in an ongoing way requires reciprocity in the form of feedback from peers and leaders, and validation that their work is driving action.

In exchange for the visibility of their work, contributors ask for evidence that their insights are being put to use. Therefore, our team is currently working on features for the insights database aimed at enabling social engagement and activity tracking for insights — features that allow contributors to respond to comments and questions, track the number of views, and understand how their work is being used to stimulate more engagement.

By directly citing insights, bodies of work, and authors, we validate the hard work of contributors and create a currency system for insights and their application. At the end of the day, everyone would like to see how their work is making an impact on the decisions of the company.

The future of insights is now

This project originally envisioned a technical problem with a technical solution: a repository for insights. As we worked, however, we quickly came to understand that a piece of software — while vital — could only be one component of our solution. We realized that, after all, insights are more than just pieces of information. They are applied knowledge: created, developed, and improved upon by people and their networks together, not in isolation.

We needed to respect all aspects of this relationship to ensure quality and actionable insights. We built our strategy on three inextricable components:

  • Partners: Empower a cross-functional, global insights network
  • Process: Establish processes to generate and leverage insights to identify gaps and inform priorities
  • Product: Enable access to insights for ongoing projects and planning through an open, self-serve tool

By building a tool that facilitates dynamic, sustained interaction between insights creators and insights consumers, we create something much richer than a simple repository of static information. Just as our network of people becomes self-reinforcing, so too does our network of insights, as creators and consumers cross-reference and make connections across products and categories. By establishing a process with engaged users, Kaleidoscope becomes the best kind of informational network: one that teaches us what it is in real-time.

As that network evolves, so will this project and our Insights Platform function. Because, ultimately, what is a kaleidoscope but a mechanism for the simultaneous creation and appreciation of beautiful patterns? As we continue to create those insights and learn from each other, Kaleidoscope will continue to change and grow.

About the author:

Etienne Fang leads the Insights Platform in UX Research at Uber. She is dedicated to increasing the impact of insights globally and cross-functionally at the company to enhance collaboration. With nearly two decades of experience in research, design, and strategy, Etienne thrives at the convergence of these disciplines to create new value for organizations. In and outside of work, she is passionate about people and the power of their stories to inspire change.

Connect with her on LinkedIn. Follow her passion project about women around the world: “Having It All Project” on the web www.having-it-all.org and on Instagram @having.it.all

Special thanks to:

Molly Stevens

Insights Platform team: Hallie Saber, Isabelle Elias

Kaleidoscope Designers: Charlie Waite, Stephanie Brisendine

Kaleidoscope Engineers: Benjamin Booth, Chris Duarte, Chris Hale, Eugene Sosnov, Garrik Sturges, Mo Kouli

Collaborators: Bob Cowherd, Jeb Burchenal, Pomi Tefera, Steph Tung

Supporters: Manik Gupta, Michael Gough, Mike Shoemaker

If you are interested in joining our team, please check out the current job listings.



Etienne Fang
Uber Design

I am a human-centered research strategist who thrives at the intersection of insights and storytelling. www.having-it-all.org