Doximity is the largest community of healthcare professionals in the country - with over 80% of U.S. doctors and 50% of all NPs and physician assistants as verified members.
Doximity's goal is to afford doctors more time to dedicate to patient care. To achieve this, Doximity offers a range of tools that streamline career navigation, networking, scheduling, telehealth, and various other tasks incidental to providing healthcare services.
Doximity builds and operates many products and services, the team used Jupyter notebooks as their main analytics tool to deeply explore their data to find insights to further improve their offerings. Unfortunately, Jupyter notebooks primarily serve Python-savvy data developers, and many SQL-first data practitioners at Doximity were unable to glean value from the notebooks.
Jupyter notebooks were great for the Python enabled folks to solve ad-hoc questions, but the SQL-first team members could not get them set up properly and found it difficult to collaborate. Doximity also found that when presenting findings to stakeholders they couldn't share the notebooks directly, instead they had to screenshot different charts inside the notebook to share.
They wanted a cloud-hosted notebook where they could write SQL or Python and then create a shareable dashboard. They wanted all of their team members to have a full and rich notebook solution no matter where their expertise fell between the two languages.
Doximity chose Hex to empower everyone on their team to do exploratory data analysis and easily share their work securely.
SQL & Python
Hex made querying Snowflake, their Data Cloud, simple and secure. Doximity’s team can write ad-hoc queries and get answers fast without having to pre-model the data. When they end up creating metrics in Hex that they want to define consistently across the organization they can then build those inside of Snowflake for future users to query directly in Hex.
In each notebook, Hex also pre-installs and manages popular Python packages eliminating the overhead of getting up and running with a notebook. Every query returns a dataframe so the Doximity team can transform, visualize, and prototype scripts with SQL or Python.
Ad Hoc to Operational
Exploratory data analysis often creates insights that should be reusable and consistently available to an organization. At Doximity, explorations of their Snowflake usage and costs and Airflow jobs have turned into critical dashboards for monitoring their data stack.
When someone calls the Doximity support line for help it's essential for their support team to have up-to-date information about the user's dialer usage. The support team uses a link to a Hex app which is pre-filtered with the relevant data of the person calling for help. This process makes it super fast for a support team member to get the information they need to help the caller.
Many of Doximity’s goals and projects end up being initially explored through Hex ad-hoc and then become productionalized via a Hex app with quality and performance metrics of the project.
All of these Hex dashboards are easily shared with a link to everyone in the organization. No access request needed. Making it very easy to quickly share with people who might otherwise have to go through a slow access request process.
Hex also allows Doximity to set security controls at the data, logic, and app level, this ensures no one is accessing data they aren’t supposed to throughout the analysis and sharing process.
With Hex and Snowflake, Doximity was able to completely transition off of Jupyter Notebooks. This made it possible to enable over 60 people at Doximity to ask and answer questions with data, plus the rest of the company can now learn from these data-driven insights with viewer access to notebooks and visualizations that leverage the scalability of their Snowflake Data Cloud. In addition to making it easy to explore data in a notebook format, Hex's components feature has allowed Doximity to re-write less code. Doximity already has over 1000 projects on Hex and loves the search and admin features that make it easy to curate their organization’s knowledge.
Erik Selin, Engineering Director Data Platform