Sellforte Experiments: Geo Lift Deep Dive Guide

Sellforte Geo Lift helps you measure the causal impact of a marketing activity by comparing treated regions against a synthetic control. It is designed for teams that want to understand whether a channel, campaign, or budget change drove incremental business outcomes rather than just observed correlations.

Geo Lift is part of Sellforte Experiments, which brings experiment analysis, AI summaries, and MMM calibration into one workflow.

What is a Geo Lift test?

A Geo Lift test is a geo-based incrementality experiment. It measures impact by:

Selecting one or more test regions
Applying a treatment, such as increasing, decreasing, or pausing spend
Comparing actual performance against a synthetic control

The synthetic control is built from similar regions and estimates what would likely have happened without the treatment. The difference between actual performance and that counterfactual represents the incremental impact of the marketing activity.

Sellforte’s Geo Lift methodology uses Bayesian synthetic control to estimate this effect, giving teams a practical way to quantify incremental impact with statistical rigor.

Why use Geo Lift in Sellforte?

Geo Lift is useful when you want to validate whether a channel or media change is truly incremental.

With Sellforte Geo Lift, you can:

Measure the causal impact of a marketing intervention
Quantify incremental sales, margin, and profit
Estimate iROAS and confidence
Review results in a visual, shareable format
Store experiments in one centralized library
Use high-quality experiment results to help calibrate your MMM over time

This is especially valuable when you want stronger evidence for budget decisions and a more reliable connection between experimentation and ongoing Marketing Mix Modeling.

When to use Geo Lift

Geo Lift is usually the right choice when:

A campaign ran in some regions but not others
You changed spend in selected markets
You paused activity in certain areas
You have enough regional variation in both media and KPI data
You want a credible treated-versus-untreated comparison

In practice, Geo Lift works best when you have a clear treatment, a clean separation between test and control regions, and a KPI that can be tracked consistently over time.

Before you start: what makes a strong Geo Lift design?

A well-designed experiment matters more than simply running more experiments.

A strong Geo Lift setup usually includes:

Clearly defined test and control regions
Sufficient treatment intensity
Enough duration to observe impact
Good pre-test comparability between groups
Reliable KPI and media data at the selected geographic level

If the test is weakly designed, the result may still run, but it will be less useful for decision-making and less suitable for MMM calibration.

How to create a Geo Lift test in Sellforte

1. Open the Experiments module

From the left navigation menu, go to Experiments.

This opens the Experiment Library, where all experiments are stored in one centralized view.

2. Create a new experiment

Click Create new experiment.

3. Select GeoLift

Choose GeoLift as the experiment type.

4. Define the experiment setup

Fill in the required setup details:

Experiment name
Use a structured naming convention, such as channel + region + date.
Group by
Choose the geographic aggregation level, such as region, state, or market.
Test groups
Select the treated regions.
Control groups
Select the regions that will be used to construct the synthetic control.
Ad platform / tested channel
Select the channel being tested.
Start and end date
Define the intervention period.

5. Run the analysis

Click Analyze to generate the result.

Sellforte then processes the experiment and opens the results dashboard, where the outputs are presented in a consistent structure with visual reporting and AI-generated interpretation.

What you can see in the Experiment Library

The Experiment Library gives you a centralized overview of your experiments.

Typical columns include:

Experiment name
Experiment type
Status
Start and end dates
Spend
Test iROAS
Confidence
Last modified date

From here, you can:

Open any experiment
Sort by columns
Compare tests at a glance
Review confidence levels
Create a new experiment

This makes it easier to manage Geo Lift analysis as an ongoing part of your measurement workflow instead of treating each test as a one-off project.

How to read the results dashboard

Once analysis is complete, Sellforte opens the experiment results page.

The top section typically includes:

Spend
Incremental sales
Test iROAS
Sales lift
Confidence
R-squared
Number of observations
Number of predictors

These headline metrics give you a quick read on both business impact and result quality.

Key metrics explained

Incremental sales

Incremental sales show the estimated business impact caused by the treatment. This is the difference between observed performance and the counterfactual.

Test iROAS

Test iROAS shows the incremental return generated by the treatment relative to incremental spend.

For example, if test iROAS is 3.69, the estimate implies that each additional €1 of spend generated €3.69 in incremental revenue during the test window.

Confidence

Confidence indicates the probability that the treatment effect is positive.

For example:

97% confidence means there is a 97% probability that the effect is positive

Higher confidence generally means stronger evidence that the treatment created positive incremental impact.

R-squared

R-squared measures how well the synthetic control matched the test region before the intervention.

Closer to 1 indicates a stronger pre-test fit
Lower values suggest weaker comparability

A stronger pre-test fit generally makes the result more reliable.

Number of observations and predictors

These help describe the structure of the analysis and the data supporting the estimate.

The AI Summary

Sellforte includes AI summaries to help turn experiment outputs into usable business insight.

The AI Summary typically highlights:

The test design
The synthetic control composition
The estimated incremental effect
Credible intervals
A business-level interpretation of the result

This is often the fastest way to understand what happened and communicate the result internally. It is especially useful when you need a concise explanation for marketing, finance, or leadership stakeholders.

How to interpret the graphs

Target KPI vs. scaled counterfactual

This chart compares:

Actual performance
Counterfactual prediction
The highlighted test period

It helps you see whether the treated region diverged from what would likely have happened without the intervention.

For example, in a go-dark test, if actual performance falls below the counterfactual during the intervention period, that suggests the paused channel had been driving incremental value.

Cumulative treatment effect

This chart shows how incremental impact accumulates over time.

A typical pattern is:

Flat before treatment
Divergence during the test window
Final value showing total estimated incremental effect

This is often the clearest visual view of the experiment’s magnitude.

Media investment vs. counterfactual

This chart shows actual versus predicted spend behavior.

It helps confirm that:

The treatment was applied as intended
Spend changed materially during the test
The intervention is visible in the data

iROAS distribution

This distribution shows the range of likely iROAS outcomes.

Typical elements include:

Mean iROAS
Lower and upper credible interval bounds
Probability mass above zero

If most of the distribution sits above zero, that indicates stronger evidence of positive incrementality.

Test and control groups

The Test and Control Groups section explains how the synthetic control was constructed.

It typically shows:

Test region or regions
Control regions
Weight assigned to each control region

These control weights:

Sum to 1
Reflect similarity to the treated region
Indicate how much each control market contributes to the counterfactual estimate

This makes the experiment more transparent and easier to interpret.

Comments and collaboration

Sellforte also supports collaboration around experiment interpretation.

You can use comments to:

Add internal notes
Record business context
Document final conclusions
Capture follow-up actions

This is a simple but important way to preserve the learning from each experiment, especially when results are later used to support planning or MMM calibration.

Using Geo Lift results to calibrate MMM

One of the most valuable uses of Geo Lift is that high-quality experiments can help improve your MMM.

Experiment results can contribute:

Real-world ROI anchors
Informative priors in Bayesian modeling
Additional evidence for channel-level incrementality

This helps connect controlled testing with broader measurement and planning. Sellforte’s calibration approach is based on the idea that MMM becomes more reliable when it is informed by high-quality external evidence, including incrementality tests.

Important: not every experiment should be used for calibration

Only use high-quality Geo Lift experiments for MMM calibration.

Good indicators include:

Strong R-squared
High confidence
Clear treatment intensity
Adequate test duration
Clean and interpretable design

More experiments do not automatically improve MMM. Better experiments do.

Best practices for Geo Lift tests

To get the most useful results:

Use clearly defined test and control regions
Make sure treatment intensity is meaningful enough to detect
Run the test long enough to observe impact
Review pre-test matching carefully
Check whether the synthetic control is credible before acting on the result
Document findings and decisions directly in the experiment
Prioritize Geo Lift for channels where incrementality learning could materially affect spend decisions