Database connection: connecting Looker data to Sisu’s UI

In our partnership with Google’s Looker, the main goal was to connect our two analytics tools to provide a more robust analytics experience. However, before the user could begin to reap the benefits of this new UI, they first had to connect their data source to our tool. Connecting their databases to Sisu had the potential to be a very lengthy, technical, and confusing process that would have likely resulted in high drop off rates. My goals for this flow were to remove any unnecessary steps, to make the technical steps easier to complete, and, to validate user steps along the way.

Keep onboarding simple, easy

To start, I designed two onboarding flows, one from Snowflake and the other from BigQuery. These two platforms covered 90% of users’ databases and provided a sufficient starting point to onboard new Sisu users.

The onboarding flow is the first experience users have with a new product so I felt that it was critical to ensure that the process was easy to use and understand, simple to troubleshoot, and easy to complete.

From a design perspective, my goals were to build trust with users early through:

  • Reducing onboarding friction and confusion in what is generally a complicated setup and experience

  • Enabling a successful connection in one go (vs. the user getting stuck, frustrated, and abandoning the flow) and maybe coming back later

While the flow is incredibly simple (only three inputs and creating a scratchpad schema), this did require making strategic design decisions.

  • We brought in the “Test Connection” button to ensure that the credentials were valid from the start. In reality, the user could have gone through the rest of the flow without without validating their credentials. However, if they made an error, they would have had to have gone back after getting through the flow to correct them. We made the assumption that if the user got through the flow, only to realize it didn’t work, that they’d abandon the flow and ultimately, not connect to Sisu. To prevent this risk, we brought in the button to ensure a frictionless continuation through the onboarding.

  • Creating the scratchpad schema requires leaving the Sisu environment to complete. Our goal was to make it as simple as possible to execute this last step and to bring the user back to onboarding flow to finish the connection process.

    • The easiest way to solve for this was to bring the commands directly the user (vs. having them look for them elsewhere). I also provided them with supporting text on where to go to create the scratchpad schema, and enabled copy + paste of the code to prevent typos.

Use progressive disclosure to guide the user down the path of least resistance

To encourage users to focus on the correct areas of the flow and to submit the right information in the right order, I reduced the overall amount of information on the screen and I used progressive disclosure to reveal specific and relevant information and inputs at the right time. To do this, I:

  • Brought in checks/validation points along the way (i.e., the user couldn’t proceed if their login credentials were incorrect)

  • Allowed for an easy copy/paste of the commands they needed to create a scratchpad schema

  • Brought in banner notifications to close the feedback loop between the user’s interactions and its outcome.