The debate surrounding probabilistic versus deterministic cross-device tracking is nothing new. But with the rapidly evolving online landscape and technological capabilities, and with customers increasingly engaging across multiple devices, brands and agencies should be having a different conversation: They need to look beyond which targeting method to use and determine how they can best identify customers for more personal and relevant engagement while continuing to maintain consumers' privacy. This is especially true as the issues of trust and scale have recently been raised concerning first-party data companies, such as Facebook.
Utilizing deterministic methods, based on some form of specific identifying data on a consumer (commonly logins, registration data, physical addresses and sometimes offline customer data or IDs, etc.), a company can determine who a specific user is. But this method has caused major issues for first-party data companies. Facebook faces this dilemma, since not only is its data limited to the information from just its platform, that data also contain very specific user information, creating a privacy issue when used with third parties.
Probabilistic methods, on the other hand, use a data science approach to take a variety of signals across multiple channels to build user profiles with anonymous data, and can increase scale by predicting behaviors of users based on similar known users. With the multitude of devices and touch points for companies to compile data from, probabilistic methods have evolved past the point of just tracking cookies. With privacy being a huge concern in today's data-driven world, probabilistic methods allow companies to create holistic customer profiles and target their desired customer segments without requiring the use of identifying information.
Along with specific, yet anonymous, customer data under their belts, companies now need to ask the question of how to best leverage the data they have access to. Marketers need to be focused on scale and accuracy, and the key to this is the amount of data a company has access to and how it's used.
Reaching Customers at Scale
The issues of both trust and scale are repeatedly a challenge in advertising technology. Probabilistic methods are able to overcome both these challenges and help companies reach consumers at scale.
In order to scale successfully, a company must be able to recognize, and tie across devices, virtually every digital consumer across a range of digital identifiers. This can be achieved through a sophisticated mix of technology and analytics that addresses all the consumer touch points and devices being used now. While this requires putting additional effort and assets into either an internal data analytics arm of the company or a probabilistic agency partner, both can help condense and analyze online and offline data to understand customer behavior and patterns, and offer superior scale for campaigns. When done correctly, probabilistic targeting methods are highly effective in terms of scale—reaching more of the right people.
Making Campaigns Accurate
The importance of accuracy when it comes to customer-driven campaigns cannot be emphasized enough, and the rate of accuracy goes hand-in-hand with the maturity of a company along the probabilistic spectrum. The more data inputs a company has access to (both in quantity and diversity), the more sophisticated it can get with its targeting and accuracy. Both proprietary and third-party data will give companies a complete consumer view. But for accurate campaigns, a company cannot just splice together third-party audience segments bought on a marketplace. It needs to have a fully integrated data platform so there's no loss of consumer data or inability to see a consumer over time.
For example, a social media platform can offer tremendous customer data and insights into its users, but it's limited to its own universe. In addition to pulling data from multiple data sets, it's equally important for companies to filter out bad data and avoid ad fraud. Skilled analytics teams can identify and filter out this bad data, use multiple technologies to reach and understand users, and tie online and offline actions.
As a company matures along the probabilistic spectrum and leverages an increased quantity of data sources, it will actually have more privacy advantages since it won't be using first-party identifiable information and can keep users anonymous.
Focusing on the customer and offering them a personalized and unique experience is the No. 1 goal that should be driving targeting and tracking methods. But privacy also needs to be factored into these campaign decisions. With the amount of data available today, probabilistic methods allow companies to continue running highly targeted and personal campaigns while ensuring the anonymity required by consumers today.