Notion’s modern data stack includes the Snowflake Data Cloud, Amplitude, Fivetran, Spark, Tableau, and of course Hex. Snowflake is our single source of truth; everything feeds into Snowflake and Snowflake feeds out to everything. We use Spark and dbt for running pipelines.
Before bringing in Hex and Tableau, our users started complaining that their notebook tool wasn’t good enough, their querying tool wasn’t good enough, and their visualization tool wasn't good enough. So we moved to Hex for querying plus notebooks and Tableau for visualization.
We wanted something that was low overhead for our small team and flexible enough to handle anything from exploratory analysis to complex data science workflows. A key requirement was finding a tool that would support non-SQL capabilities like Spark. Hex was a great fit because of its low overhead, and it worked right out of the box without much setup or maintenance.
We expected to use Hex in traditional notebook ways, but we were surprised to see that our most active users were actually on the GTM side, including our CX and sales teams. Our CX team uses published Hex apps as a UI to handle common tasks, like pulling details about a particular customer. On the sales side, our SMB team uses Hex to combine data from Salesforce and Snowflake to calculate targeted offers for prospects.
Outside of our GTM team, Hex is being used by our data scientists for things like churn analysis, expansion analysis, and ML workflows. And this year we want to get to the point where data can feed back into the product, starting with better search. This team has been using Hex on top of Snowpark to push down SQL and Python compute directly into Snowflake allowing them to work with and transform large amounts of data quickly. Due to Hex’s flexibility and integration with Snowflake we are able to support a wide range of technical capability in one tool.
It’s great to see our data scientists and GTM team interacting in the same data tool. Our vision for Notion’s data team is that anyone on the team, regardless of technical proficiency, is comfortable using data to answer their own questions and Hex helps us enable that.
A year ago, we didn’t understand the shape of the Notion business – is Notion primarily a personal users’ business? Is Notion primarily a teams business which expands as teams grow? Where does Notion fit in the SaaS economy? We fundamentally didn't understand that, and that made it way harder to strategically figure out what we wanted to do in the next year. That was the number one lift – understanding the shape of Notion’s business – and Hex made a big impact.
Abhishek Modi, Software Engineer
Notion