Navigation
data.world is a cloud-native data catalog built on a knowledge graph. Our platform gives organizations more transparency into their data stack and helps them simplify data discovery and access. The tools within the platform are created with both technical and business-minded folks in mind, providing easy-to-use features for data discovery, governance, enrichment, and much more.
Our platform was initially designed for the data hobbyist community, but since expanding our user base to enterprise customers, we’ve had to revisit some of our core features such as navigation. This was a year-long initiative that spanned many different workstreams with an overarching goal to change our navigation model to accommodate our vast range of customer use cases. Our user base includes Fortune 100 companies, non-profits, and data hobbyists, so their concerns and data catalogs vary significantly.
Company
data [dot] world
In collaboration with
Design & Engineering leadership, Engineers, Customer Success, Solution Engineering
Biggest challenge
Balancing vastly different use cases
What was exciting
Being able to do thorough concept validation
THE PROBLEM
The inability to represent our customers’ assortment of resources created a confusing and inconsistent navigation experience throughout the platform.
This problem was visible in our core navigation features: global navigation, organization profile, and home page.
The current global navigation divided all resources into two predetermined buckets: data and analysis. This was an assumption we made a year ago that we’ve since learned was not an ideal sorting mechanism. It forced everything, even “Tableau dimensions” and “DBT models” into these two categories causing unnecessary confusion. Users didn’t understand how their resources were sorted.
Since expanding our customer base to enterprise users, we’ve had to find ways to optimize their use cases with the existing community-oriented pages. One of those features was the organization’s landing page. The existing platform landing page was a Facebook-like activity feed that allowed data hobbyists to see recent activity on their projects. This wasn’t the starting point our enterprise customers wanted to see, so we created the “organization landing page” as an interim solution. We quickly outgrew this temporary fix as we realized that we needed a unified page that (1) works across both community and enterprise, (2) could be configured for different customer needs, and (3) would display the right information as a starting point.
Addressing these problems meant that we could also sunset or combine similar pages. This would result in more consistent page layouts that are easier for our internal teams to manage during development.
THE PROCESS
We dedicated a significant amount of time on low fidelity designs in order to efficiently collaborate with other departments that have a better understanding of our customer use cases.
In the early stages of our process, we spent many cycles iterating on the design in low fidelity and set aside any visual and micro-interaction concerns for the future. We mapped out how our design would work for customers that were most representative of our user base. We discussed the implications and challenges with our Customer Success and Solutions teams as a first user research proxy. This allowed us to get feedback on our direction early on and address any potential concerns beforehand.
Once we had aligned on a clear concept, we tested prototypes with users to validate our assumptions and solutions.
One of the major changes we originally wanted to introduce was the ability to “switch” navigation content based on the level of access users had in an organization. Since this required our users to relearn how to navigate through our platform, we made it a major part of our user research focus. We created interactive prototypes for our participants to see, and asked a lot of questions about how they expected it to work or what they expected to see.
In the end, we got consistent feedback that this “switching navigation” concept was not an intuitive feature. However, our participants had indicated that their original navigation issues were addressed through our new organization profile and home page experience. So the constructive feedback on navigation made it easy for us to cut the complexity out of the delivery plan, saving a significant portion of engineering resources.
Introduced new roll-out processes and patterns.
Since these changes would be seen across our entire platform and directly affect all our users, we had to be very intentional about how we rolled them out. Our previous processes mainly consisted of quietly AB testing and notifying the Customer Success team. That would not be sufficient this time, so we built out in-app patterns and interactions to preemptively notify our users about the changes.
First, we had banners that allowed them to preview the new version, identify which pages that would be removed, and share feedback. We introduced a new visual pattern for identifying “preview” features and used Mixpanel to keep track of user interactions with those components. This gave us a lot of quantitative feedback, such as how many users were aware of the change and how many users kept the new version on. These roll-out patterns ended up working out well and served as a good template for our later roll-out plans.
THE SOLUTION
Our overall strategy was to keep the navigation features simple and have defaults for all personas to easily understand; but we also recognized the need to allow granular configurations for our enterprise customers at the content level.
Global navigation
The design of this simple navigation is sorted by ‘most frequently visited’, and it combines organization-level pages with personalized pages. Enterprise users will likely jump straight to “Organization” which shows a high-level overview of what is in their organization’s catalog. If they are in multiple organizations, as most admin personas are, they will see an interstitial that allows them to choose which organization to jump into. For our community users, “Organization” will only show up if they are members of any. Otherwise, the top item is “Discover”, which takes them to an empty state of the search page that surfaces all the resources they have access to.
Organization profile
As noted above, we had created the organization landing page as an interim solution. Our new design consolidated similar pages, resulting in the “new organization profile”. This ensured that all links that referenced the organization were redirected to a single, unified page as a home base for the organization. It has a markdown section for admins to personalize messaging to their end users as well as different ways to browse and discover what is available in their catalog.
Home page
Lastly, we have the home page, where we made the most drastic changes. Originally, it was an activity feed of passive information. Our enterprise users wanted to know what needed their immediate attention and links to jump to their most frequented pages. So we changed the home page to be a widget-based landing page that could flex based on the user’s experience in data.world. If they had access request notifications or comments on resources they owned, that would be at the top of the page. If they are a part of any organization, those links would be at the top as well. Information that was less important, but required occasional attention, such as the activity feed, would be closer to the bottom.
THE IMPACT
“It feels like everything that I need is right in front of me.” - User research participant
In this year-long process, we were able to deliver a well-rounded product story and consistent experience to our user base with updates to the navigation, organization profile, and home page. These new updates have made it easier for our Sales team to showcase our platform configurability and more efficient for our Implementation team to set up the platform for our customers. The user experience elegantly builds on top of itself as enterprise customer needs become more complicated, but the carefully crafted decisions we made for the default use case work great for our community users as well. Out of all the work I've been involved with at data.world, I am most proud of this series of projects that touch the core UX of our platform!