WHY NON-CONSENT BIAS SHOULD BE TOP OF MIND FOR MARKETERS

What is non-consent bias?

Non-consent bias is a phenomenon where individuals who opt out of data collection systematically differ from those who consent. This can lead to biased or inaccurate results because the (consented) sample may not represent the larger population. As a result, non-consent bias poses a critical risk of distorted data and results for advertisers. 

What is the impact on advertisers? 

A recent study by Exactag on 2022 data from over 25 advertisers highlights the critical nature of non-consent bias. It found that non-consented users have higher basket values (+ 5%), convert faster (-33% time to conversion), and require fewer touchpoints (-20%).

Based on our study, neglecting these findings around non-consented users means every campaign appears less efficient and effective. For example, the insights for tactical planning, such as campaign timing, will consequently lead to wrong decisions. And with rising opt-out rates, the risk gets bigger for advertisers. That's why they must think about their strategies, even five years after the launch of GDPR.

What can be done to minimize the risks?

Advertisers need to optimize their consent strategies and banners to minimize the risks associated with non-consent bias. But another vital instrument is often overlooked: carefully choosing the analytical approach to take to tracking and analysis under this challenging situation. Why? Because when you rely on biased data, all analytical instruments will fail, whether you're using descriptive statistics, incrementality tests, media-mix models, or attribution.

To address the challenge of balancing privacy and performance measurement, there are – among others – three instruments available.

1. Highly aggregated data
Some analytical marketing instruments never use user-level data, including marketing-mix-models (MMM). Typically, these instruments use channel-level data. This is why market players initially chose MMM as the cornerstone of their performance measurement. But unfortunately, many projects undertaken failed to ensure that the digital data used was unbiased. Moreover, advertisers adopting this approach must live with the shortfallings around MMM, especially its limited flexibility, and granularity.

2. Extrapolation
Other market players and advertisers choose to run analysis on consented data only and then extrapolate the results from this subset and apply them to all users. Some of them use pragmatic approaches like corrections factors for extrapolation. Others, like Google's consent mode, apply probabilistic models to create a database of all users. The biggest problem here is trust. It's essential to trust stable trends, but the nature of black box solutions and non-transparent models means it’s hard for advertisers to have confidence in them.

Privacy experts also devalue some extrapolative approaches as they try to model individual user behaviors despite an opt-out.

3. Micro-aggregates
Another approach that brings together privacy and performance measurements is based on micro-aggregates. It's a well-established approach in offline marketing. For example, Deutsche Post has been using so-called micro-cells for years. In digital marketing, tracking micro-aggregates is a promising alternative as it solves the main disadvantages of the approaches mentioned above. This approach entirely relies on deterministic rather than probabilistic data, and micro-aggregates hold more real information than highly aggregated data.

The challenge with this approach is that it requires intensive knowledge and resources to further develop existing instruments in a way that can integrate new types of data.

The Exactag Solution

Exactag has developed a hybrid tracking approach that collects user-level data for consented users and group-level data (micro-aggregates) for non-consented users. This delivers integrated results for advertisers from the channel level down to the creative level. It compares consented and non-consented traffic, providing detailed customer journey data for consented users. And when it comes to privacy and legislation, Exactag's group-level tracking does not process personal data, so it doesn't fall within the scope of the GDPR.

When users request a service, their consent is also not needed to collect data under Section 25(2) of the TTDSG. For reach measurement, if the cookie is set according to strict rules, it fulfils the requirements of this, as confirmed by the Bavarian Data Protection Authority in its 11th activity report in 2021. France, Italy, and the Netherlands agree with this, and we strictly follow these guidelines in everything we do.

If you’re struggling with non-consent bias and looking to improve the effectiveness of your campaigns by minimizing it, then please get in touch.

Contact us.