Which factors explain changes in demand?
Rank the variables that contribute predictive signal across time, market, channel or product.
Demand Driver Analysis · Predictive + Causal
Know what moves demand and which levers you can actually change.
Separate useful predictive signals from credible intervention levers so commercial teams can distinguish what explains demand from what they can act on.
Questions this decision model answers
Start with the commercial or operational choice.
Which drivers are actionable?
Distinguish controllable commercial variables from macro, seasonal or contextual signals.
Which levers are likely to change the outcome?
Use causal methods where the decision requires intervention evidence rather than correlation.
Where should commercial effort move?
Combine driver evidence with feasible actions, expected impact and business constraints.
From data to decision
Move from explanation to intervention evidence.
Not every variable that improves a forecast is a business lever. The modelling approach should match the question being asked.
Demand + candidate drivers
Combine sales or volume with price, distribution, promotions, media, weather, macro, channel and other relevant variables.
Measure explanatory signal
Estimate which variables help explain and predict changes in demand.
Test intervention questions
Apply causal methods where the business needs to estimate the effect of changing a controllable variable.
Prioritise credible levers
Return the strongest supported levers, expected direction or effect and the evidence limitations.
Decision boundary
A strong predictor can still be a bad intervention target.
Weather, seasonality or macro indicators can improve demand prediction without being controllable. A commercial variable can correlate with demand because it was changed in response to demand. The decision question is therefore not only what predicts demand? but what can we change and expect the outcome to move?
Signals
Variables that improve explanation or forecasting.
Levers
Variables the business can actually change.
Interventions
Levers with evidence that changing them is likely to change the outcome.
Related decision problems
Follow the decision chain.
Related guide
Causal machine learning
Estimate intervention effects rather than relying on correlation.
Related guidePromotion incrementality
See a concrete causal demand-driver problem.
Related guideCPG & Beverages Decision Pack
Apply demand drivers to pricing, promotion, distribution and commercial action.
Direct answers
Questions buyers ask about this decision.
Is demand driver analysis the same as feature importance?
No. Feature importance can show which variables help a predictive model. It does not by itself establish that changing a variable will change demand.
Can external variables be included?
Yes. Weather, macroeconomic indicators, calendar effects or market variables can be useful predictive context even when they are not controllable.
When do we need causal methods?
When the business wants to estimate the effect of an intervention, such as changing price, promotion, distribution or media investment, rather than only explain historical variation.
What does the output look like?
Outputs can include ranked predictive drivers, actionable levers, causal effect estimates where supported, uncertainty and recommended areas for intervention or testing.
Decision IntelligenceFrom signal and prediction to a governed business action.AutoMLAutomate repetitive model-building work without losing business context.Causal Machine LearningSeparate drivers from correlations and estimate intervention effects.Predictive AnalyticsForecast demand, score customers and anticipate business outcomes.Decision OptimisationCompare feasible actions and rank them by expected business impact.Forecasting vs OptimisationSee where prediction ends and the business decision begins.Decision APIDeliver recommendations, expected impact and evidence into workflows.Embedded Decision IntelligenceAdd a decision layer to ERP, CRM, BI, workflow and vertical software.Build vs BuyA practical framework for deciding whether to build, buy or embed.
Bring the decision, constraints and data you already have.
We will structure the model around the action your team needs to take and the evidence required to support it.