What’s Next After the Cookie Goes Away? Incrementality

The question burning on every marketer’s mind right now is how we approach measurement in an uncertain future. With the cookie going away in 2023, we can anticipate an impact on marketing vanity metrics. However, if the right solutions are implemented, it won’t make an impact on the most important metric: sales conversions.

While there are several different approaches to take in the post-privacy world — from probabilistic attribution to media mix modeling — incrementality is, and will become, an even more important tool in the marketer’s measurement toolbox.

Moving forward, geo-based incrementality in particular will enable marketers to understand the impact marketing strategies have on customers’ decisions.

Addressing attribution through incrementality 

Historically, deterministic attribution models have allowed advertisers and vendors to directly connect user conversion touchpoints with user media touchpoints across different marketing channels. However, the absence of a unique user identifier caused by Safari ITP, iOS 14.5, and the forthcoming elimination of the 3rd-party cookie in Google Chrome makes this connection difficult, if not impossible, to establish. While probabilistic attribution will continue, marketers will need additional measurement tools to move forward — such as incrementality. 

The general concept is as old as the ‘controlled experiment’, which dates back to the 18th century. This was used to determine a subject’s response to two context-driven options, and in turn, which option is more effective. It’s commonly used in fields including healthcare, sociology, and of course marketing. The process is as follows: 

• Two randomized groups are formed: one is called the control and the other test
• The test group receives a change (e.g., a specific amount of ad spend) while leaving the control group unimpacted (e.g., no ad spend change)
• As the experiment time passes, observed outcomes are compared between the test and control groups to evaluate the overall impact of the independent variable

While the concept is simple, implementation requires nuance. 

Taking incrementality to the next level 

As marketers strive to figure out how to implement new forms of measurement into their workflows, many will have to not only prioritize incrementality but also a geo-based version of the tactic. This means tracking tactics such as serving ads, ad spend, and outcome metrics by geography. In order to do this, marketers need to pick geographic regions and randomly assign some of them as control groups and the remaining areas as the test group (e.g., applying DMAs). From here, the test group receives increased ad spend while the control group does not, allowing for outcomes at the end of the experiment to determine the impact of incremental ad spend. 

However, when executing this strategy, there are many detailed considerations marketers must have in mind, including: 

• Randomizing geography choices: It’s helpful to organize geos by size when selecting control and test groups
• Baseline measurement and seasonality: For many advertisers, there is already existing brand equity and product demand that drives baseline and seasonal outcomes – it’s important to use statistical models like linear regression to remove such effects from the analysis
• Existing ad spend: Some geographies may already have ongoing marketing and advertising investment – it’s critical to only consider the incremental spend to measure incremental lift

Geo-based incrementality is an attribution model-agnostic approach that can be used for almost any measurable outcome from online transactions to offline sales.

Why incremental measurement is a solution to privacy changes 

Some marketers might argue that privacy changes make it more challenging to track the incremental success of campaigns.

A privacy-centric world does make it more difficult for marketers to track and follow individual users across the customer journey. A major use case here is in attribution modeling where it is critical to understand the end-to-end touchpoints of a single consumer through the purchase funnel. 

By contrast, most incrementality (i.e., control vs. exposed testing) measurement is done with aggregate data. In the specific case of geo-based incrementality, we can measure impact across cohorts of users in a specific location without the need to understand individual consumer behavior.

Finding the right solution 

Geo-based incrementality may not be the exact right tool for everyone. There are other options such as ghost bidding and deterministic lift studies that are all worth exploring. 

While all approaches have pros and cons, it’s crucial to consider them each through the lens of a specific brand’s needs. However, as we continue to approach a new privacy-centric era, brands must make action plans, and geo-based incrementality provides a flexible and agile approach that can benefit many brands.

Huanlei Ni is SVP of Product, Marketing, and Data, at Goodway Group.