Build a test worth running.
Match comparable geographies, estimate power, and choose a duration that can detect the effect you care about.
Open-source incrementality platform
Measure what your
marketing actually creates.
Design geo-lift tests, build Bayesian counterfactuals, and turn experimental evidence into a clear decision.
MODELBayesian counterfactual
SIGNALGeo time-series
OUTPUTPosterior causal lift
ENGINEPyMC / Python
EXPERIMENT 04 / COMPLETE
LONDON — Q2 BRAND CAMPAIGN
MODEL LIVE
POSTERIOR LIFT
+14.8%
95% credible interval+8.2% — +21.1%
DECISIONScaleEvidence tier A
ObservedCounterfactual
μ 0.148σ 0.033P(LIFT>0) 99.8%
01 APR15 APR01 MAY15 MAY01 JUN
CAUSAL MEASUREMENT / 01
Attribution reports who clicked.
OpenLift measures what changed.MODEL STACKPYMC 5ARVIZPANDASSCIKIT-LEARNSTREAMLITPYTHON 3.9—3.12
01From a hard question
From a hard question
to a defensible answer.
OpenLift brings experiment design, causal measurement, and business decisions into one transparent workflow.
Estimate the lift that attribution misses.
Use Bayesian synthetic controls to model what would have happened without the campaign.
-5%
+25%14.8%
Move from probability to action.
Get evidence strength, economics, limitations, and a recommendation your team can act on.
SCALE HOLD CUT RETEST
02One workflow.
One workflow.
Four clear steps.
Bring your data
Upload geo time-series data or connect Google Sheets, Google Ads, and Meta Ads.
Design the test
Match markets, calculate MDE, and validate pre-period fit before spend goes live.
Estimate lift
Model the counterfactual, quantify uncertainty, and diagnose experiment quality.
Make the call
Review impact, economics, and the next best action—then save it to your scorecard.
$ git clone https://github.com/daramolaben10/openlift.git $ cd openlift && pip install -r requirements.txt $ streamlit run app.py OpenLift is ready. Create, measure, and review your first geo-lift experiment.Installation guide
03
CREATOR / MAINTAINER
DB—001
CREATED BY DARAMOLA BEN
Building clearer tools
for harder decisions.
Daramola Ben is the economist and open-source builder behind OpenLift. His work connects rigorous causal measurement with practical decisions teams can defend.
MSc Applied EconomicsBSc Economics & Econometrics
</>
BUILT IN THE OPENYour measurement logic
Your measurement logic
shouldn't be a black box.
OpenLift is MIT-licensed, inspectable, and built for teams that want to understand—and improve—the methods behind their decisions.
04Understand the method,
Understand the method,
not just the metric.
Technical field guides grounded in OpenLift’s implementation and experiment methodology.
FIELD GUIDE 01
Marketing incrementality
What incrementality measures, where attribution stops, and how causal evidence changes budget decisions.
Read the guide FIELD GUIDE 02Geo-lift testing
How to choose markets, validate controls, plan test windows, and avoid contamination.
Read the guide FIELD GUIDE 03Bayesian synthetic control
How OpenLift builds the counterfactual, represents uncertainty, and calculates posterior lift.
Read the guide05OpenLift,
OpenLift,
explained clearly.
Direct answers for teams evaluating open-source incrementality measurement.
What is marketing incrementality?+
Marketing incrementality measures the outcomes caused by marketing—not simply the conversions associated with an ad interaction. It compares observed performance with a credible estimate of what would have happened without the campaign.
How does OpenLift measure geo-lift?+
OpenLift matches test and control markets, validates pre-period fit, and uses Bayesian synthetic control to estimate the counterfactual. It reports posterior lift, credible intervals, evidence strength, and economic impact.
What data does OpenLift need?+
At minimum, OpenLift needs daily or weekly geo-level time-series data with a date, geography, and outcome such as revenue, orders, or conversions. Spend, channel, treatment, and creative fields add richer analysis.
Is OpenLift free and open source?+
Yes. OpenLift is MIT-licensed and available on GitHub. Teams can inspect the methodology, run the platform locally, contribute improvements, and adapt it to their measurement workflow.