Atlas SQL Interface
Atlas SQL Interface
product designer @ MongoDB
product designer @ MongoDB
To provide our secondary personas, analysts, first class access to their MongoDB Atlas data from their coveted SQL-based tools.
goal.
I was the lead product designer of the Atlas SQL Interface for about a year and lead various feature introduction and improvement efforts.
my role.
DURATION
THE TEAM
March 2022 - November 2023
1 Product Manager, 3 Engineers
PROJECT TYPE
product design, stakeholder management
In this project, design priorities came primarily from product management, business requirements and insights into user needs. These included:
business needs.
The key design opportunities from business and engineering needs thus included:
Make Atlas SQL easy to get started with and highly visible to our target analytic users.
Transition current BI Connector users seamlessly to Atlas SQL through clear deprecation communication.
Hide and automate information on Data Federation creation to simplify the Atlas SQL process.
1.
2.
3.
4.
Provide further SQL schema management customization for power analytics users.
Design priorities were also balanced out by engineering efficiency and feasibility needs:
engineering needs.
While MongoDB Atlas is a powerhouse for data storage and management, there is a growing need to target the data analytics space. In order to do so, we want to scale our existing tooling - Atlas SQL. How might we make enabling Atlas SQL visible and frictionless for analytic users to activate and manage in order to connect MongoDB data to external SQL tools?
problem space.
Increase investment into analytics users by simplifying the first touch experience for Atlas SQL.
Grandfather out the static payment plan and introduce an improved and scale by usage one.
Connect the UI with our user’s main programmatic interface experiences through further SQL schema management of downstream data.
1.
2.
3.
Decouple Atlas SQL from Data Federation manual set up to reduce needed engineering resources for large background data federation instances.
Deprecate the legacy BI Connector tool to redirect engineering and support resources to a more scalable analytics tool.
1.
2.
With these design opportunities informed by business and engineering needs, an MVP high level user flow navigating how to easily enable and inform users of the new Atlas SQL experience was needed.
user flow.
I worked closely with my PM and engineers to iterate over feedback from both a business and feasibility perspective.
lofi exploration and hifi prototyping.
Once a solid iteration was approved through internal review, I ran unmoderated user testing with external MongoDB users to quantify the discoverability and ease of use of Atlas SQL.
user testing.
7
30 min unmoderated interviews with external MongoDB users
4
UI task walkthroughs on finding and configuring Atlas SQL
Quantitative survey responses to each of the tasks:
* 1 = very difficult, 7 = very easy
Qualitative user sentiments between the old manual experience versus the new Atlas SQL quick start set up:
Users found setting up Atlas SQL through standard Data Federation creation as a bit more “cumbersome” and highly “specific”.
Users found the new quick start Atlas SQL creation experience as “easy”.
The key insights from user testing included:
Visibility into Atlas analytics tools has improved.
While Atlas SQL standard creation is much more manual, it is still provided for users that want increased personalization through the CLI.
Current BI Connector users are being correctly informed and transitioned to the new tool.
1.
2.
3.
4.
Simple versus power users should have slightly different experiences - the latter has more awareness and access to further management tools.
After the last round of iteration based on user testing feedback, the solution space was becoming more clear. The biggest MVP requirement was to decrease the barrier to entry for Atlas SQL by bringing the access to the home page, automating and abstracting background Data Federation set up happening, and allowing for advanced configuration options for experienced users.
solution 1: decoupling from Data Federation.
We then identified all touch points for Atlas SQL awareness and provided messaging that emphasized proper solutions to accommodate the change.
After developing a quick start way to activate Atlas SQL, our secondary concern was to direct our current BI Connector customers towards this new option. I worked closely with my PM to create a deprecation timeline for when in the user journey to surface what kind of messaging. Sensitivity was needed to ensure the correct users were being informed in advance.
solution 2: deprecating BI Connector.
Finally, after the top two priorities were address and implemented, a third round of product improvement occurred. In this sprint, adding in additional Atlas SQL schema management capabilities within the UI was prioritized.
solution 3: adding SQL schema management.
An A/B test experiment was ran between these two treatments:
A: Variant with original manual Data Federation instance creation.
B: Variant with the quick start Atlas SQL experience entrance in the connect modal and cluster card.
success tracking.
+64%
SQL Awareness: # of organizations that clicked on the “Connect” CTAs
+39%
SQL Acquisitions: # of organizations that ran an Atlas SQL query
+2000%
Average SQL queries per organization
Due to the statistically significant success of the above experiment, the quick start Atlas SQL experience was officially implemented. Segment tracking was also implement to monitor it’s usage in comparison with the deprecation of BI Connector:
+53%
Quarter over quarter: Total Atlas SQL meaningful usage organization
+42%
Quarter over quarter: Total Atlas SQL weekly active organizations
-34%
Quarter over quarter: Total BI Connector weekly active organizations
I teamed up with user research afterwards to further understand usage and pain points/ opportunities of Atlas SQL in connection with visualization tools.
further research and design.
4
1 hour moderated interviews with external MongoDB Atlas SQL users
3
interview priorities - ecosystem, data visualization, and analytics tools
4
1 hour moderated interviews with internal technical support experts
The key insights from user research and potential future Atlas SQL projects include:
Users appreciated the quick start option and that Data Federation was hidden from the Atlas SQL experience.
Many users have pre-selected an external data visualization tool but would prefer to have various data viewing options live in MongoDB Atlas instead.
Users wish to build everything in one ecosystem and hope that Atlas SQL could have more extensive configuration capabilities and a dedicated analytics home within MongoDB.
I then worked with design systems to investigate and standardize deprecation patterns and guidelines after transitioning away from BI Connector and other products within the Atlas suite.
1.
2.
3.
After internal and external design audits, product team interviews, and data analysis and prioritization, these became the finalized deprecation guidelines:
Design systems collaborated heavily with product design in order to update our banner component to include a CTA for deprecation use cases:
14
internal audits of MongoDB deprecating products
2
analysis and prioritization sessions
7
30 min moderated interviews with product teams.
challenges.
learnings.
The biggest learning from the challenges above was to definitely prepare design kickoffs with as much understanding of product goals and engineering timelines as I could. Then, I would use the kickoff time for aligning on design goals, confirming the stakeholder groups, and clearing up any ambiguous questions I had which led to much more seamless iterations later on.
It was very exciting to work with a team of highly collaborative and communicative PMs and engineers. While it led to quick feedback turnaround and consistent synchronous check-ins, it also opened the door to a multitude of conflicting opinions and perspectives that I had to balance and prioritize design-wise. Furthermore, since the Atlas analytics space was moving fast with the emergence of Atlas SQL and the deprecation of BI Connector, there were several tight sprints and timelines to keep in mind - descoping was constantly necessary.