Understanding Cookieless Attribution and Measurement in Digital Advertising
Why wait for the industry to decide on cookieless attribution and measurement? Here’s a method that already beats the old ways
The discussion around technology replacement for third-party cookies has so far largely been centered around tracking and targeting. But a key element that should equally be in the spotlight, and seems to have been largely forgotten, is how new ways to reach consumers will be measured, attributed and optimized.
To date, digital advertising has used cross-site and cross-app tracking techniques for targeting. But without an accurate measurement and attribution model in place, the open web would be broken: publishers could not attribute conversions and g******* for them, brands would be unable to understand ROI and ROAS of their digital media and the rest of the ad-tech supply chain would break down.
In simple terms, attribution allows us to understand whether the action of serving an advert in a particular environment or site led to its desired outcome, such as the purchase of a product or service on another site.
The two main cookieless attribution proposal types under discussion across the industry are event-level reporting and aggregated reporting: the former is client-side, favored by Safari and Chrome, that associates an ad click or impression view with conversion data; the latter a server-side solution to reveal total counts across larger groups of people.
But while some progress has been made, the industry is still far from reaching a robust model similar to what cookies allowed us to measure – or even a consensus on the approach and how the necessary computation should be done.
So, What Can I Do to Measure the Effectiveness of My Campaigns Without Third-Party Cookies?
With data expertise and advanced machine learning techniques, combined with existing statistical models, it is possible to come up with a measurement offer that breaks the silos, factors in channels, seasonality and other factors and allows advertisers to determine what an optimal allocation of investment should look like.
For this to work in cookieless attribution, we need to turn to marketing mix modelling (MMM).
This is data-driven statistical analysis that quantifies the incremental sales impact and ROI of all online and offline marketing activities. It doesn’t require individual or log level information, it works for all sectors and is highly customizable. It also enables actionable decision-making and continuous innovation.
Using a modern, always-on version of the marketing mix model – one that uses statistics and machine learning to balance tactical and strategic outputs – we can minimize analysis bias and quantify the incremental impact on sales and ROI for all online and offline marketing initiatives.
This in turn allows us to analyze:
Share of Spend versus Share of Effect
By calculating incremental sales driven by each channel, we are able to evaluate ROI and effectiveness across all media channels and platforms.
Effects of Seasonality on the Cookieless Attribution Model
By weighing actual vs predicted response, we can capture the effects of seasonality on campaigns and in turn predict its future effects.
AdStock Decay Rate
Advertising Adstock represents the decay effect of awareness in the minds of the consumers after exposure to advertising. By understanding the decay rate for each channel, we can detect the long-term effect of advertising on consumer purchase behavior.
These results give us the tools to continually optimize budget allocation. By assessing the effectiveness of all channels in multiple, highly accurate ways, we can clearly see which channels are over- and under-performing and adjust spend shares to maximize the effectiveness of a campaign.
Private, customizable and actionable
A method like this is privacy-friendly by default, highly customizable across industry verticals, and provides actionable outputs that brands can use to inform their own decision-making and deploy in their overall strategy.
Also, not only is it the most effective of all the cookieless options available, but it has substantial advantages over attribution models that were in use even before cookies began to be rolled back.
Because even with the aid of third-party cookies, attribution models were always inconsistent. The last-click attribution model has been much overused as a ‘source of truth’ whilst not taking into account other elements, including ad formats, creative, seasonality and the contribution of online and offline channels such as TV or radio.
All of this gives us plenty of reasons to think that the industry may not have been measuring effectiveness properly all along. And new methods like this give us great hope that the future is not simply one of finding workarounds for old cookie-fueled methods, but of making real and meaningful advances on all fronts.
Comments are closed.