A practical five-step framework for making better organizational decisions with data. Designed to be applied to any significant decision under uncertainty.
Step 1 — Define the Decision
- We have stated the decision in a single specific sentence (not "how are we doing?")
- We have identified who has the authority and responsibility to make this decision
- We have set a deadline for the decision
- We have identified the status quo — what happens if no decision is made?
Step 2 — Identify What Data Would Help
- We have specified what data would change our decision if it were different
- We have avoided looking at data before specifying what we need (to reduce confirmation bias)
- We have identified the most relevant data available, not just the most accessible data
- We have identified what data we wish we had but do not
Step 3 — Evaluate Data Honestly
- We have assessed the completeness of the data (are there known gaps?)
- We have assessed the representativeness (does it cover all relevant populations?)
- We have assessed the recency (does it reflect current conditions?)
- We have documented known limitations that affect how confident we should be
Step 4 — Apply Expected Value Thinking
- We have identified the plausible range of outcomes, not just a single expected outcome
- We have estimated the probability of each scenario honestly, not optimistically
- We have calculated the expected value across scenarios
- We have identified which scenarios we are most uncertain about and why
Step 5 — Document and Review
- We have documented the decision, the data used and the rationale
- We have set a review date to evaluate the decision's outcome
- We have assigned someone to monitor the relevant outcomes
- We have a plan for what to do if the decision produces unexpected outcomes