How it works
The method answers one question rigorously: what would your KPI have done if the intervention had never happened?
The counterfactual idea
Your metric always moves — seasonality, trends, market noise. A simple before/after comparison confuses these movements with the effect of your intervention. Instead, we build a statistical model of what your KPI would have looked like without the intervention, learned from its own history and from control variables that were not affected. The gap between reality and this counterfactual is the estimated effect — delivered with confidence intervals, not just a single number.
Not a before/after
The model accounts for trend and seasonality, so ordinary fluctuations are not mistaken for impact.
Uncertainty included
Every estimate comes with a credible range, so you know how confident to be.
Cumulative effect
See not just the daily effect, but the total accumulated impact since the intervention.
What you receive
We deliver CSV files containing your results, together with a PowerPoint presentation or a PDF report showing the findings.
CSV result fields
response- Actual KPI.
point.predpoint.pred.lowerpoint.pred.upper- Predicted KPI without the intervention, with its 95% interval.
point.effectpoint.effect.lowerpoint.effect.upper- Daily effect = actual − predicted, with its 95% interval.
cum.responsecum.predcum.pred.lowercum.pred.upper- Running totals of the actual and predicted KPI, with the predicted total’s 95% interval.
cum.effectcum.effect.lowercum.effect.upper- Cumulative effect since the start of the evaluation window, with its 95% interval.
What we need from you
At least one year of data
A daily or weekly time series of your KPI covering at least 12 months before the intervention, so the model can learn its normal behavior.
The KPI to measure
The metric on which you want to measure the effect: revenue, conversions, visits, churn — whatever matters to your decision.
Control variables
Related series that explain your KPI's usual variation but were not affected by the intervention. You can provide them, or when possible we can research suitable ones for you.