Data governance: an 8-step program
Imagine you join a company to run a mobile app. You want to build a sign-up funnel to see where users drop off. You log in to your favorite user analytics tool, and there are 4 signup events. Which one is right? You ask around, and everyone is using a different one. How can you pull numbers accurately? Is everyone being misled by the data?
Every organization needs a sound data governance process or program to maintain and optimize the health (and trustworthiness) of their data. It is the only way to ensure you get the answers you need. In this article, Meghan Swidler, Global Partner Operations at Mixpanel, walks us through a step-by-step guide that you can use and modify for use within your own organization. Previously a Senior Implementation Consultant on the Mixpanel team, Meghan has helped many of Mixpanel’s large clients build out their company-wide analytics strategy and implement Mixpanel.
After successfully completing your initial analytics implementation, you know that you’ve established the proper foundation to measure progress against your current business metrics–a great first step in the right direction. However, as your product and business objectives continue to grow and evolve over time, those metrics are bound to change. Your data schema should reflect this so that you can continue to measure progress against key metrics and make data-informed business decisions.
1. Select a data governance owner or team
If you belong to a smaller, more nimble organization, select a primary data governance owner, who may be the owner of the initial analytics implementation. You should also choose a backup data governance owner who can quickly and easily step into the lead role if the primary owner cannot take responsibility.
If you belong to a larger organization, create a data governance team or governing council with a lead from each functional business unit leveraging this data (e.g., Product, Marketing, Analytics, Data Science) so that each team is well represented. This will also help to break down silos that may exist across teams.
2. Create a centralized implementation spec for your product
Once your data governance owner or team is in place, make sure that you have a shared implementation spec to document new events and properties. You should already have one if you worked through your initial analytics implementation, but if you don’t, feel free to copy our implementation spec template (and to reference our industry-specific implementation specs linked in the same article).
3. Document new events and properties
Before launching a new product feature (whether in alpha, beta, or to all users), the product manager responsible for the launch should establish the right metrics (to hold him/herself accountable) and submit a request to the data governance owner or team. The data governance owner or team can then build out the events and properties required to measure progress against these metrics.
(Need help building out a proper data schema? Check out Phase 1 of the Guide to Mixpanel Implementation and/or reach out to our world-class Professional Services team to guide you through this process.)
4. Review these new events and properties
When documentation is complete, the data governance owner or team can review these new events and properties with the product manager. If you have a designated Mixpanel Professional Services account team, you can also have them review and provide feedback on the new events and properties.
Either way, get sign-off that the events and properties actually map back to quantifiable business metrics and are ready to implement!
5. Implement the new phase
The data governance owner or team can now relay to the technical lead(s) that they can proceed with development, translating the events and properties in the implementation spec into triggers within the product’s source code. The process here may vary drastically depending on internal processes, but typically starts with the creation of an engineering ticket by the data governance owner or team.
6. Conduct proper QA (in your development environment) before deploying to production
The data governance process doesn’t end once the track calls have been added to your product’s source code. Now’s the time to engage in quality assurance processes to ensure that the data being collected is accurate and aligns perfectly with the new events and properties documented in your implementation spec.
Once you’ve confirmed that the data is accurate, you can deploy to production!
7. Document your event and property definitions
If you use Mixpanel, the data governance owner or team should add descriptions of all new events and properties in Lexicon and organize data for clarity and discoverability. This will help everyone across your organization (who may or may not have been part of the implementation process) understand what data is being collected, which will in turn empower them to run analyses within Mixpanel and make data-informed business decisions.
And last but not least,
Step 8: Repeat Steps 1 through 7!
I hope that this step-by-step guide helps you implement the data governance process that maintains the health of your implementation and helps you measure progress against key metrics so you can make data-informed business decisions and build products that people love.