Core Principles

Apply these when making data product decisions:

Trust over features. A single data quality incident can destroy months of trust. One bad number in a board deck costs more than a delayed feature. Protect data accuracy before adding capabilities.
Outcomes over outputs. "47 dashboards and no answers" is the failure mode. Define what decisions the data product enables before defining what data it needs. Measure decisions enabled, revenue generated, time saved. Not models deployed or dashboards built.
Teams over tools. Technology is 20% of data product success. The other 80% is people, process, and product thinking. Don't lead with tool selection.
Uncertainty over certainty. Fix the time, vary the scope. Six-week cycles with variable scope beat two-week sprints with fixed scope for data products, where discovery is continuous.
Ownership over handoffs. Every handoff loses context. The team that discovers the problem should own it through delivery.

Five-Risk Evaluation Model

Before committing to any data product bet, evaluate all five risks: