Metrics for a product company tell us where we are and help us decide where to go. Metrics fall on a spectrum from high-level to low-level.

The way I like to think about metrics is grouped by business, product, service and system.

Service in this context is the thing built, maintained and run by Software Developers to implement the product. It is not the public-facing service users consume or pay for.

Similarly, a system in this context is where the service is running. These days it is in the cloud like AWS, but it could also be a server in one’s own data center.

Metric owners

This grouping of metrics corresponds to distinct owners:

• Executives own business metrics
• Product managers own product metrics
• Developers own service metrics
• Cloud provider owns system metrics

Metric examples

• Business metric: Revenue
• Product metric: Conversion rate
• Service metric: Checkout count
• System metric: CPU usage

Metric cadence

Metrics on this spectrum also correspond to slower to real-time time scale.

• Business metric: Quarterly
• Product metric: Daily
• Service metric: Minute
• System metric: Real-time

These time scales are as slow as they can be without impacting our decision-making ability. We need to see the EC2 metrics in real-time to figure out which server is dying. How a new feature affects the conversion rate needs a few days to soak.

Metric method

Metrics are implemented with different methods.

• Business metric: Accounting
• Product metric: Events
• Service metric: Counters
• System metric: Native

Event tracking can be implemented by services like Mixpanel. Counters can be implemented with AWS CloudWatch custom metrics.

Do you want to see the full spectrum in full action? You’re in luck, Battlefy is hiring.