At the heart of our innovative products lies a commitment to data-informed development. This approach is not just about referencing data but understanding and integrating it at every crucial point of our product development process. Discover how our team approaches data-informed product development.

From the inception of a new feature to ensuring top-notch data quality, our data scientist shares a step-by-step look into our process.

1. Tracking user behavior is not an add-on

Your data analyst should be in the loop right from the get-go, rather than being clued in after a feature’s rollout.

How we do it:

1. A new feature is designed and approved for implementation.
2. The product manager informs a data analyst about it.
3. The data analyst creates a tracking draft.
4. Developers integrate the feature and tracking.
5. The data analyst gives the green light post-testing.

2. Detailed documentation is worth it

Don’t keep everything in your head, scattered over Slack messages or Jira tickets. Everyone, including your future self, will be thankful.

How we do it:

1. Entire tracking is defined in Avo (or a well-organized Google Sheet).
2. Avo is used as a single source of truth.
3. Every tracking change is defined directly in Avo. Jira tasks only point to it.
4. If a bug is discovered, the event is updated with a bug status and a description.

3. Simplicity is key

Do not track every single action. Ask questions, and keep it simple. For example:

1. What are the key objects and actions we want to track?
2. Why are they important? 
3. How will these insights drive action?
4. What will happen if we do not get these insights?

If you do not have clear answers, it is likely that the tracking is unnecessary.

4. Taking control of your data quality is key

Data quality should be the No. 1 priority of every data team. If people do not trust the data, it is worthless.

How we do it:

1. Full awareness of the tracking status is achieved by implementing the previous 3 steps.
2. In each release, both new and old tracking is tested. The more important the tracking, the higher the detail and frequency of the tests.
3. Regular tests are run using dbt, Avo inspector and custom Python scripts to check the health of our tracking.
4. It might sound obvious, but it is often overlooked: Fix the bugs.

The post The Road to Data-Informed Product Development appeared first on MochaLabs.