Home » Insights » Why MMM and Extrapolations are not the answer to the GDPR

Heavy on Data. Light on cookies

One concern for advertisers was how this loss of non-consented data could lead to distorted performance statistics. Results from analytical techniques relying on user-level data (e.g. individual customer journeys) were viewed with scepticism as the underlying data volumes declined. And to address this issue, alternative analytical approaches like media mix models and incrementality tests were discussed as potential replacements.

But what’s critical is that any solution adheres to three essential areas to provide a future-proof data collection approach: it complies with all privacy legislation; it’s complete or representative of 100% of the data; it allows analysis across multiple dimensions like channel, campaign, publisher and creative.

So how do the options stack up to these requirements?

  1. Extrapolation
    Advertisers can work with opt-in data only and use it to extrapolate out to 100% of sales and customers. It’s easy to apply this, it’s privacy compliant, and it provides a complete dataset. However, opt-in users behave differently from opt-out users – something we’ve learnt from our customers, data and industry research. This makes extrapolation risky as it impacts the reliability and validity of results. And in today’s uncertain environment, trying to apply correction factors is difficult due to unstable user patterns.
  2. Media mix modeling
    Advertisers can use MMM-like econometric models to analyze the impact of their channels, campaigns, publishers, etc. With a wealth of providers in the market offering standardized models based on practical algorithms, it’s easy to access these services. And this approach is privacy compliant and representative due to the highly aggregated data used. But aggregated data lacks depth. Because it doesn’t provide granular data that’s refreshed daily, it can’t deliver the insights needed to guide tactical and operational activities. This means many stakeholders won’t be supported, including media buyers.
  3. Workarounds for user identification
    Advertisers can establish workarounds to overcome missing consent. Fingerprinting, eTag and other techniques are options. But while offering rich, meaningful data at scale, they’re not privacy compliant. Brands run the risk of compromising their reputation. Indeed this applies to any technological approach that allows a single user to be identified. And any workarounds attempted will be equally non-compliant.

Taking a different approach

That’s why we strive to get the maximum value out of the available data under these new conditions. And we don’t abandon opt-out data. We accept that modern attribution must deal with heterogeneous data sets, so we created a new data set for non-consented events. It’s collected anonymously at a grouped level using a well-established compliant segmentation approach from offline marketing. And it’s used by Deutsche Post and Acxiom.

The advantage is the data’s not distorted. Instead, it is meaningful, deep, and granular, allowing for complete customer journey analysis. And by comparing opt-in and opt-out patterns, it delivers additional insights to improve marketing effectiveness. It also complies with all legal requirements relating to the GDPR and TTDSG and is legally verified by ePrivacy GmbH.

That’s why we strive to get the maximum value out of the available data under these new conditions. And we don’t abandon opt-out data. We accept that modern attribution must deal with heterogeneous data sets, so we created a new data set for non-consented events. It’s collected anonymously at a grouped level using a well-established compliant segmentation approach from offline marketing. And it’s used by Deutsche Post and Acxiom.