All marketers really want to know is, “Did my advertising campaign make an impact? And how much?” In a world of walled gardens, of media plans that are divided across measurement platforms, and of web browser cookie opt-out policies … getting these answers isn’t a cake walk. But, it’s also far from impossible. At CvE, how can we tell what reactions are triggered when the direct observation is unavailable? Enter lift-based measurement or incrementality studies.
The concept of lift is a simple idea, but it requires a strict, scientific method approach to ascertain direct cause and effect. Lift is defined as “having an enhanced response measured against a random choice targeting.” In other words, when all else is held constant, something changes in response to a stimulus. The point of this approach is to determine the incremental impact of a marketing campaign relative to what would have happened anyway. If you shut off all advertising tomorrow, customers will still make purchases. But, how much more would they have made had the campaign been live? In the absence of a lift study, advertisers may fall victim to celebrating media metrics resulting from cookie bombing those already-high-propensity users.
What are your options? Each type of lift study seeks a baseline for comparison but from different angles.
- Pre-/Post-Lift: This is by far the simplest type of lift study, yet it is most likely to be skewed by seasonality and trends.
- Example: Measure your audience or market response for two weeks; then, run your ad campaign for two weeks and measure the change in conversions.
- Year-over-Year (YoY) Lift: This holds for seasonality and requires less data-gathering than subsequent options, yet it can be impacted by regionality and industries affected by climate changes.
- Example: Run a Cyber Weekend campaign and compare to the previous year’s Cyber Weekend campaign results.
- Control vs. Exposed Lift (aka hold out study or placebo study): This type of lift study is the gold standard in public health because it requires a random selection of the target audience to receive an ad while another random selection of the same target audience does not.
- Example: Two user ID pools of the same data-backed qualifications are identified. One is served the ad campaign and the other is served an unrelated PSA ad.
To begin, it’s important to identify the ultimate goal – your North Star. Do you want to increase awareness? Rates of trial? Sales? Foot traffic? This will guide all measurement decisions there on out to ensure all efforts are aligned to the goal. Then, identify measurable, obtainable proxies for this goal, such as transaction amounts, number of downloads, or certain correlated on-site activities. Needless to say, clicks are not proxies – they are cop-outs. The majority of conversion and online engagement has been proven time and time again to happen in the absence of clicks. Do you click on ads? Nor does 86% of the internet-browsing population.
To execute a lift study, look to the scientific method to ensure accuracy and replicability: observe, hypothesize, test, implement, repeat. You’ll observe a desired outcome, hypothesize the impact a stimulus will make, test two randomly selected samples with and in the absence of the stimulus, confirm or deny the hypothesis, and finally implement the results into strategic, budgetary and/or optimization decisions. Rinse and repeat. Ongoing test-and-learn lift studies across tactics, channels, geographies, audiences, etc., will continue to reveal incrementality on your quest to reach optimal results.
When data collection, organization and maintenance are comprehensive enough, statistical modeling options will present themselves to identify the strength of correlation across cause-and-effect relationships (how much impact does each variable have on the outcome). This is the holy grail and unlocks a world of data-driven decision-making for your future campaigns.
For better or worse, the quest for better results never ends. The ad tech world changes at a rapid pace, but in this world lives oceans of data. And it’s up to you in how you navigate these seas by following your North Star. We’d love to be your first mate. Reach out to us to talk more about our unique lift-based approach and how you can implement a lift study on your next campaign.
As senior analytics specialist at CvE, Nick guides our teams to find meaning in data. He applies an end-to-end test-and-learn philosophy to help deliver the insightful reporting and strategic-planning services CvE’s clients deserve. Nick believes numbers tell stories and good campaign strategy is a delicate blend of craft and science. An expert in the digital arena, he has spent the last 12 years working directly with regional, national and Fortune 500-level clients to provide strategic digital media advice and his take on the latest industry trends.