From dashboards to decision systems
Why the next frontier in analytics is not more reporting but better decisions�supported by integrated data, explicit decision logic and continuous performance feedback.
Read article
2 min read
Author: KeynesMoore
What if the central assumption is wrong?
A plan is often most vulnerable not to a missing decimal, but to one shared belief: demand will recover, financing will remain available, a supplier will deliver, regulation will permit the model, or customers will accept the change. Scenario analysis should expose what happens when that organising assumption fails.
Start by naming the assumption in measurable terms, including horizon and range. Build a small set of coherent alternatives around the mechanisms that could invalidate it, not arbitrary percentage shocks. Trace first-order effects and then feedback: lower volume may worsen unit economics, constrain investment, weaken service and reduce demand again.
Use sensitivity analysis to locate thresholds and reverse stress testing to work backward from an unacceptable outcome. The Bank of England�s 2025 guidance emphasises both tools: sensitivity analysis reveals how key assumptions shape results, while reverse tests identify the boundary at which risk becomes material or the business model fails.
For each scenario, specify leading evidence, decision points and feasible responses. A mitigation that requires capital, supplier capacity or regulatory approval after the shock is not yet a response plan. Test whether several teams are relying on the same scarce resource and whether management has enough lead time to act.
Do not select one scenario as a disguised new forecast. Compare strategies across the range and favour actions that preserve options, reduce irreversible exposure or remain valuable in several futures. The purpose is not to predict the surprise; it is to prevent one unexamined assumption from carrying more risk than leaders consciously intended.
Related macro
Analytics, data and decision systems
Turn data, analytics and decision systems into clearer insight, stronger choices and more effective execution.
Discover the macroEditorial overview
Articles
Planning for uncertainty with simulation and optimization
How companies can move beyond point forecasts by combining scenarios, predictive models and optimization to improve decisions under volatile conditions.
Read articleFocus
What will you do differently if the forecast is right?
Prediction creates business value only when the organisation knows which decisions and actions should change as expected outcomes change.
Which metrics actually change a decision?
A performance measure matters when it alters management attention or action, not simply because it can be reported consistently.
Strategic challenges
Most performance systems look backward
Historical reporting remains dominant even when the decisions managers face depend on drivers, scenarios and changing future conditions.
Enterprises produce data faster than they create information value
Large data estates can continue expanding while important users still recreate datasets and struggle to find reliable information.
POV
Precision is not the same as insight
A model with another decimal place is worthless if decision-makers still cannot explain what matters or what they should examine differently.
Accuracy is not the objective
A predictive model should be judged by whether it improves the decision it exists to support, not by statistical performance in isolation.
Strategic impact
Product thinking changes the unit of data investment
Reusable information assets can concentrate ownership and engineering around needs shared across multiple consumers and applications.
Optimization makes trade-offs explicit
Objectives and constraints can be represented directly, allowing competing uses of resources to be evaluated within the same analytical problem.
What we observe
Complex models often answer simple questions badly
We frequently see analytical sophistication increase while the business question, assumptions and intended decision remain poorly defined.
Governance often produces more policy than trust
We frequently see extensive frameworks while ownership remains nominal, metadata incomplete and quality problems unresolved at source.