- Tags:: 🗞️Articles, 📖 Data Science for Business, Metrics frameworks
- Author Daniel Schmidt (cofounder of DoubleLoopApp, for product teams to visualize their strategy)
- Link:: Metrics-driven product development is hard. (doubleloop.app)
- Source date:: 2021-12-13
- Finished date:: 2021-12-22
A similar article than Machine learning does not produce value for my business. Why. but expanded idea on which metrics to pick. It is kind of like a causal diagram:
To tie work to business results, you must create a graph of relationships between work and metrics. E.g., “If we optimize X in the product, then metric Y will go up. If metric Y goes up, then that improves our North Star metric. And if we increase our North Star metric, the business will make more money.”
Each link in the graph is an assumption that could be wrong or go wrong. You can form hypotheses for how work will move input metrics and for how leading indicators influence lagging indicators. But your hypotheses, of course, can prove false. And a correlation that exists at one point in time might break in the future.