Our Approach
We build microproducts by working from a real decision to a usable product. The sequence is deliberate: name the problem, find the signal, make the data reliable, then ship something people can use.
1. Identify Problem
Start with a decision worth improving.
A microproduct begins when someone repeatedly faces a choice that is slower, noisier, or riskier than it should be. Name the decision, who makes it, and what better would look like before you reach for tools or data.
2. Identify Data
Find the sources that reliably capture the signal.
Once the decision is clear, ask which data would change it. Prefer sources that are available, trustworthy, and tied to the outcome you care about. If the signal is missing or weak, the product will be too.
3. Build Pipeline
Collect, clean, and model the data end-to-end.
A notebook is not enough. Turn the raw inputs into a maintainable path: ingestion, cleaning, transformation, and models that can refresh when the decision needs them. This is where analytics engineering makes the work durable.
4. Build Microproducts
Turn the pipeline into microproducts people can easily use.
The last step closes the gap between insight and utility. Ship a focused experience, workflow, alert, or tool that puts the modeled data into someone's hands. Then learn from real use and improve.
How This Maps to the Playbook
These four steps are the conceptual arc. The rest of the playbook expands them into practice through Frame, Build, and Operate.