Finding the Right Balance with Experimentation-based Measurement
In the first ten years of the new millennium, as internet users grew from millions to billions worldwide, the rapid adoption of digital marketing ushered in the rise of personalization. User-level targeting spawned thousands of adtech startups and fueled the meteoric rise of Programmatic Advertising. Behavioral data about users got broadcasted around the internet in search of any marketers willing to spend advertising dollars to drive demand for their product.
The Impossible Balance
At its core, data-driven targeting still represents a force of significant economic good – find the demand for your product in any corner of the world, reduce transaction costs, and grow your business efficiently. The importance of this technology for small and emerging B2C brands cannot be understated.
Even Amazon was a small B2C brand in the early 2000s and grew in their early years because they had a superpower in harvesting demand from long-tail Google PPC keywords.
As with any force capable of this much economic impact, there are collateral effects. Consumers have raised concerns that they are spooked by how we are using their data. Even if it means exposure to more relevant content, consumers are wary of their identity and information being passed around the internet to be used by brands to push products. A global effort to reform how companies collect and use what they know about consumers has resulted in new regulations and increasing limitations put on the use of third-party data.
Publishers want to respect consumers’ privacy but enabling marketers to surgically target potential customers is an effective way to monetize their platforms. As for marketers, granular measurement of media performance provides invaluable insight for future investment decisions. It presents a significant challenge to balance the needs of all parties involved.
Studying the dynamics of this privacy versus performance dilemma, some interesting observations can be made to reveal a potential way forward. Studies show that consumers accept the value exchange in being served personalized advertising when they are on the publisher’s platform, consuming the publisher’s content, just not anywhere else on the internet. By corollary, publishers have permission to target the consumer on their owned and operated properties, based on signals they have gathered through the use of their site (first-party data).
Marketers generally make investment decisions at the audience cohort, tactic and ad set level, not really at the consumer-level.
User-level measurement is not actually necessary to gain an accurate insight into media performance and make informed decisions about budget allocation. A look at how search bidding platforms have operated for the past 20 years or how Facebook buying platforms operate today reveals the potential to obtain the elusive balance between privacy, targeting and performance measurement.
Reverting Back to First-Party Reporting
In the late 90s, it was relatively easy to correlate referral traffic to downstream site events. At a time when e-commerce was brand new to the human experience and barely anyone was clicking on referring links, attributing a particular ad to a purchase served as an exclusive beacon of consumer intent and incrementality. They saw an ad, they clicked, they bought something.
Fast forward 20 years and consumers are online all the time, across multiple media platforms, and they are being served ads for the same products on Facebook, YouTube, TV, search and elsewhere. Today, publishers still provide reports that correlate ad exposure to resulting consumer behavior, but the insight is limited to a single platform. It can tell you whether your targeting is working, but due to the number of additional places consumers may have encountered your ads, it does not reveal which media is working or the incremental contribution to business outcomes.
In order to optimize media investments, marketers need to know how much each media tactic is contributing to results and how much is being wasted on that audience. That is the incrementality problem. On their own, individual reports from publishers are most useful for targeting optimization but combined with controlled experiments they can reveal accurate insight into causality and incrementality – arming marketers with the data to make informed decisions, based on science.
By using targeting parameters available within publisher platforms to configure ad campaigns that serve a control treatment and a campaign treatment to similar audiences, publisher tracking and reporting can tell us the conversion rates for both the control audience and the campaign audience. The difference in those conversion rates can be interpreted as incrementality and gives the marketer accurate data to determine how to increase performance or eliminate waste on that platform.
Because each publisher platform differs in functionality, reporting and available data, the ideal approach to setting up experiments varies between Facebook, Search, Shopping, YouTube, Retargeting, OTT and so on. Once experiment designs have been refined for each platform, controlled experiments can also be run across multiple channels to extract differential performance data at campaign and audience-levels, providing the additional granular insights needed for campaign planning decisions.
Optimizing Performance While Protecting Privacy
When consumers visit and engage with content on a publisher’s properties they are establishing and consenting to a first-party relationship with that publisher. First-party publisher device graphs also happen to be the most effective mechanism for attributing conversions, and they don’t require any cookie syncing or sharing identity with third parties. It is what publishers have always done, attribution measurement on their own inventory.
Combined with strategically designed and scientifically sound experimentation, long-established first-party publisher reporting can be the new answer to an ideal balance between publisher, marketer and consumer interests.