Core Lift Guide
Core Lift: The Certified Incrementality Study to Validate Channels That Last-Click Cannot Measure
What Is Truly Needed to Measure the Effectiveness of a Channel Like TikTok When Last-Click Doesn't Work and the Budget Doesn't Justify Marketing Mix Modeling?
A properly conducted lift test is required. Today, geo-lift is the only causal measurement tool that works without cookies, without consent, and without individual tracking, and it can even be performed with free open-source tools. The real problem is not the availability of the technique, but the ability to design, execute, and interpret it so that the results stand up to the CFO. Core Lift is Core Analytics’ incrementality study: it combines the technical execution of geo-lift with expert consulting hours that certify it as defensible evidence for the Board.
The Challenge Every CMO Faces: How to Launch a New Channel and Measure It Correctly?
TikTok is the clearest example. It's the channel every marketing director knows they should test, but many companies postpone it for a very concrete reason: no one can tell the CFO how much it truly yields.
Last-click doesn’t help. Sessions close within the app, attribution windows are short, branding effect is high, and it’s not clear what the dashboard is showing. The ROAS declared by the platform is inherently optimistic and cannot be presented to the Board. Full MMM is not feasible until the channel has just launched or the overall media budget does not exceed the critical threshold.
Two options remain. Don’t test it, and lose the channel for the next two years. Or truly test it, with the method that measures causality: a lift test.
The same applies to launching a new national radio station, for metropolitan OOH, for a sports sponsorship, or for the first investment in CTV. These are all cases where the channel is strategic, the investment is significant, but traditional metrics cannot tell if it has worked.
What Is a Geo-Lift in Practice?
A geo-lift is a geographically controlled experiment. Two groups of comparable areas are selected: in one, the campaign is activated (test group), in the other, it is not (control group). The difference in sales between the two groups, net of confounding variables, is the incremental effect: sales generated by the campaign that would not have occurred otherwise.
The reference statistical methods are Synthetic Control and Causal Impact. The output is not “the campaign generated 100,000 euros,” it is “the campaign generated between 80,000 and 120,000 euros with a 90% probability.” It works across all channels, including those that cannot be measured at the user level: linear TV, radio, OOH, sponsorships, TikTok.
Why "DIY with Free Tools" Rarely Works
Open-source frameworks for lift tests exist and are publicly available. In theory, an in-house team can perform a geo-lift on its own. In practice, most tests performed without expert support produce useless results. The reasons are recurring.
Area selection.
Test and control groups must have comparable sales in the pre-period. If groups diverge before the campaign, the effect is not attributable. Selection requires matching algorithms across multiple simultaneous dimensions. Errors make the test unusable.
Lack of power analysis.
This is the step that most poorly designed tests skip. It is discovered after conclusion that the experiment lacked the statistical power to detect the expected effect: the result is “not significant,” and it’s unclear whether that means “it doesn’t work” or “we had no way to see it.” The damage is twofold: the test budget was spent, and there’s no answer.
Interpretation of results.
Even when the numbers arrive, interpreting them requires statistical expertise. Understanding if an effect is robust to sensitivity analysis, if there’s contamination between contiguous DMAs, if a local event skewed the control, if the result is generalizable outside the test perimeter. These are judgments formed over years.
A poorly executed lift test is worse than no lift test at all. It undermines the credibility of the measurement initiative with the Board and burns the experiment’s budget. This is why some companies have attempted one and never repeat it, archiving causal measurement as “not feasible for us.”
Core Lift: The Study + The Seal of Approval
Core Lift is the Core Analytics product that brings an incrementality study into the company, certified by those who have been doing causal measurement for over 15 years.
The project scope is precise. Statistical test design and hypothesis definition. Algorithmic area selection and Minimum Detectable Effect calculation. Assisted execution during test weeks, with active monitoring of contaminations and events that can skew results. Final analysis with sensitivity tests and methodological validation. Board-ready report with strategic interpretation of budget implications.
The value is not in the software, which is available to anyone. It lies in the consulting hours within the project: the expert judgment that decides if groups are comparable, that recognizes a data anomaly as a signal and not an error, that translates a credibility interval into a budget allocation recommendation.
For the CMO presenting results to the Board, Core Lift is the difference between presenting a number and presenting a signed number. For the CFO receiving it, it’s the difference between an estimate and proof.
From Core Lift to MMM: The Natural Evolution
For many strategic brands, Core Lift is the first structured step in causal measurement. The study is conducted when needed, for a specific campaign or a new channel to be validated.
It costs a fraction of a full MMM and produces defensible answers for decisions the CMO needs to make now.
Over time, the history of tests accumulates, the media budget grows, and the organization matures.
The evolution towards Marketing Mix Modeling becomes natural: priors already exist, the methodology is shared, and confidence in the numbers has been field-tested. Lift test results are used as informative priors in the Bayesian calibration of the model, and the MMM inherits the learning from the tests.
For those starting now, Core Lift is the fastest way to enter the world of causal measurement without first having to build infrastructure, historical data, and governance.
Leave a comment