We’ve come through two decades of fascination with big data in the marketing universe. When it comes to data, we’ve been taught that bigger is always better, especially if you throw AI and machine learning at it. Yet, with the rise of privacy laws around the world, the third-party cookie, which underpins so much of the user-level data in digital marketing, will soon vanish and eliminate the data that many marketing analytics solutions are built on.
A recent survey conducted by Digiday revealed that 71% of agency and brand executives worry about their lack of a clear plan of action post cookies. There are two leading options for marketers who want to maintain user-level data for measuring media effectiveness for optimization: continue with a “big data” mindset and use non-cookie based identifiers (as cookie replacements) or turn to permission-based panels.
Kantar has found that permission-based panels are the best path forward, for many reasons. Panel-based audiences provide insight into marketers' most important tool: the impact of targeting. They also deliver more accurate insights through more representative data. Furthermore, panel-based data-sets deliver insights faster, and at a lower cost than most alternatives.
You might think of panels as a relic of broadcast measurement, where once upon a time, a tiny sample was extrapolated into the entire universe. Or you might assume that cookies and their replacements are accurate tools for capturing complex digital interactions at every keystroke. The reality might surprise you. Today, much of the data marketers depend on for consumer insights, from Nielsen media measurement to governmental tracking studies used panel based methodologies that are much improved in recent decades.
If you think you know panels and cookie replacements, stay with me in this series where we examine the two approaches to a cookieless future and explore why panels will likely prove the better choice for most marketers.
The differences in approach defined
Before diving into why we have chosen to focus on permission-based panels, let’s define the differences between the two options.
The first option, cookie-replacement, focuses on finding a non-cookie ID to use as a substitute for the cookie, but still keeps the same general approach. This method has the benefit of sticking with existing ecosystem structures in place, but instead of using a third-party cookie as the persistent identifier, it uses a replacement, like an ID created by an onboarding vendor, or an ad network. Despite the promise of stability, there are many drawbacks to this type of solution, including that it is replacing one flawed model with another just like it.
The second option, using permission-based panels, focuses on gaining insights from a smaller data set, but one that has explicitly given permission to be tracked. This panel is recruited and normalized to ensure a sufficiently large, representative sample of the marketplace being examined. As you will see below, this approach that Kantar and others have chosen, has significant benefits over the cookie-replacement option. Foremost among the benefits with this approach is the ability to provide insight into the impact of audience targeting, which cookie-based approaches most often do not.
Why measuring the impact of targeting is a huge deal
Generally, user-level analytical solutions build their universe based on who was exposed to media within the campaigns. They then look at what sequence of events or other media interactions increased the probability to convert and then fractionally attribute the conversion to those media exposures.
The problem with this approach is that it eliminates the ability to control for the impact of targeting. Targeting is one of the most useful tools in the marketer’s arsenal. Think of how much more effective targeting has become on the modern web with rich audience profiles compared to the old days of television or magazines where one might buy audiences of “men 18 - 34” or “women who cook.” Today, marketers successfully target to micro-segments that include demographics, geographics, implied and stated interests, past purchase behavior, etc. Removing this lever is a major drawback for big data solutions that only build the universe for analysis based on exposed users.
Persuadable audiences and appropriate frequency
In the analog TV world, it was historically taught to aim for a frequency of three to all audiences. In the world of addressable media, with so much control over delivery, what is optimal frequency?
Modern media planners focus on “Goldilocks” audiences. These are the persuadable consumers, with just the right amount of interest. Delivering too many ads to your existing fans is a waste of ad dollars. They’re already engaged with you and readily receive your messages – they’re too hot. Equally wasteful is spending dollars speaking to audiences that are either not in-market, not fans of your product, or otherwise closed to your message - they’re too cold. Real incremental gains in market share and revenue come from focusing optimal frequency with persuadable audiences, those that are not-too-hot and not-too-cold, but instead, just right, like Goldilocks’ porridge.
Without including exposed and un-exposed users in your sample (which is generally only possible with a panel-based solution) you can never accurately understand the impact of targeting, and without this insight, marketers are flying blind and guessing about the impact of their most powerful tool.
As marketers forge a path forward in a world without cookies, they are faced with a choice: Find a replacement for cookies, or look to other solutions, like permission-based panel data. We’ve defined the choice and outlined key limitations for insights on targeting that cookie-replacement solutions will face. In our next article, we will explore the impact of Selection Bias, the complexity of managing big data solutions for marketing analytics, and the fundamental question of privacy.