MoonPay Webinar Q&A: Your Metric Tree Questions Answered

In our recent webinar with MoonPay, we discussed how the company has successfully implemented metrics trees to bring greater clarity and solve big problems.

The webinar featured some great insights from Emily Loh, MoonPay's Director of Data, alongside one of her Senior Data Scientists, Matthew Reuvers.

But one of the best parts of the webinar? The questions from you, the audience.

So, we extracted all the questions asked and listed them out in a simple Q&A format below. Enjoy!

Note: Some of the answers may have been paraphrased and condensed for ease of reading. We've included some video clips in the article, but you can watch the full webinar to get the full, unedited answers.

How did you structure your metric tree? Did you have lots of iterations?

Matthew Reuvers: Initially, we aimed for simplicity to ensure stakeholders could easily understand. Although we experimented with various splits, we avoided overly granular splits, such as by country, due to complexity. Instead, we created adaptable structures allowing quick filtering by regions when necessary.

Did you build your metric tree using a bottom-up approach, and how common are your metrics across the business?

Emily Loh: We used a bottom-up approach, starting with data team operational needs. The metric tree primarily serves the data team but translates well across the organization. Roughly 80% of our metrics are common business metrics to maintain clarity and accessibility.

What happens after building the metric tree? How do you gain organizational buy-in?

Emily Loh: Initially, excitement was high when we introduced the tree, but active engagement required effort from our team. The data team had to proactively drive conversations around metrics, becoming strategic partners rather than reactive support. This approach significantly enhanced organizational clarity and engagement.

Have you used your metrics for predictive analytics or opportunity sizing?

Emily Loh: Yes, the bottom rungs of our metric tree contain input fields that adjust outcomes for predictive modeling. This allows us to project the impact of improving specific metrics, enhancing decision-making through clear, data-driven scenarios.

How do you handle extensive data, like blockchain or crypto data, given your space?

Matthew Reuvers: Utilizing platforms like Count enabled efficient integration of complex blockchain data, previously difficult to visualize in traditional BI tools. This allowed us to leverage detailed transactional visibility, significantly improving insights and strategic opportunities.

Can you provide a real-life example of using your metric tree effectively?

Matthew Reuvers: Recently, revenue sharply dropped in a specific region. The metric tree allowed rapid identification of the exact problem area (e.g., new customer conversion rates), significantly speeding up the investigation and resolution process compared to past scenarios.

How far down the metric tree do you typically go? Do you create specific trees for different products or segments?

Emily Loh: We designed our high-level metric tree intentionally broad to avoid overwhelming stakeholders with excessive detail. Specific areas, like marketing, can have dedicated, detailed metric trees. However, we carefully manage granularity to maintain clarity and usability.

How do you balance data self-service with maintaining organizational control and clarity?

Emily Loh: Balancing self-service and organizational control is critical. We've strategically limited the main metric tree's depth and granularity to maintain clear narratives and prevent confusion. Detailed, domain-specific metric trees are managed by respective data scientists to address deeper, tactical issues effectively without overwhelming stakeholders.