In Song of Myself 51 Walt Whitman famously wrote “I am large, I contain multitudes.” In the nearly 200 years since, people have been analyzing its meaning. What multitudes did Whitman contain? Some say the phrase is a statement on human complexities and contradictions, that each of us is defined by a vast collection of experiences.
We are not the same person from one day to the next. We meet new people. We like new and different things. We form new relationships, sever old ones. We move, change jobs and go through different life stages. Whitman of course lived before the age of the internet, mass media and the smartphone, so by his 19th century standards we carry an even greater number of multitudes, and have far more outlets for expressing ourselves.
Trying to unravel a person’s complexities is the job of brands intent on delivering a personalized customer experience (CX), which is possible only if the brand knows something about a customer. It is the reason brands deploy customer data platforms (CDPs) to create a unified customer profile that is the foundation for profitability differentiating one customer from another through personalization.
CDPs, like people, also contain multitudes in that one is different from the next. How a CDP handles identity resolution is a key difference that will ultimately determine the strength of the unified profile, i.e., the better the process of searching, analyzing and linking customer signals across disparate data sources, the better – and more trustworthy – the profile.
Types of Identity Resolution
Some CDP vendors offer identity resolution as a black box function, meaning the entire match, merge and identity stitching process is inflexible. Even if the process uses both deterministic and probabilistic matching, the issue for marketers and business users of the unified customer profile is that have no insight into why records – various signals – were or were not matched, merged or split.
Still other CDP vendors might claim to offer identity resolution, but a close look reveals that they are referring to a simple deterministic matching process – using a common identifier such as email to link various devices to the same customer.
And still other CDP vendors outsource identity resolution altogether, claiming their primary job is to collate customer data across various sources. In their telling, identity resolution is done by bouncing customer data off a third-party reference file, offering little more than a match in time. A major limitation with this method is that reference files change keys with every iteration, making it virtually impossible to gain an understanding of a customer over time, or to see how a customer proceeds through a customer journey.
Robust Identity Resolution: Transparency & Tunability
Accurate, dependable and trustworthy identity resolution, by contrast, should be completely transparent as to why a match, merge or split was made or not made. And rather than identity resolution as a black box service, tunable identity resolution allows marketers and users to adjust resolution at the individual and household levels across multiple cases, accurately mapping a customer’s personal and/or business relationships.
As an example of how a marketer may want to control the details, consider a scenario in which multiple employees of a company share access to a purchasing portal. For B2B marketing, it may be necessary to interact with an employee at an individual level in some circumstances and as a shared account holder in others. Similarly, tighter identity resolution may be a requirement of a healthcare provider sending PHI, where a looser standard may suffice for sending general information about flu shot availability.
Or, as resolution levels pertain to householding, consider a regional bank or financial services firm with an interest in determining the centers of influence when an account holder has multiple accounts, with multiple linked physical addresses. Understanding the dynamics of a household becomes important for many reasons, particularly with the potential sharing of sensitive financial information and knowing which products to offer.
A retailer, too, will want to know which family member is responsible for a browsing session when using a shared home device. In a household with a mother, father and two college kids, is it the mom purchasing bedding and linens, and why is there a new shipping address? Which family member was on the website the day before looking at blenders?
The use of probabilistic matching techniques will help determine household dynamics, but transparency and tunability are key components for providing marketers with confidence that their campaigns are targeting the right audience. If probabilistic matching determines that the mom is buying bedding for her daughter’s college dorm room to be shipped to that new address, a marketer will likely not be using that address to send offers, but may want to link it to the daughter’s unified profile for use in a different campaign.
The intertwining of transparency and tunability are key to ensure that identity resolution levels may be adjusted to best optimize the intended business outcome, to protect PII/PHI or to ensure that – for any customer or household engagement – a marketer has the proper contextual understanding of the entity.
For more on how the Redpoint CDP handles identity resolution, click here to join Redpoint VP of Product Management Beth Scagnoli and Redpoint VP of Engineering Kris Tomes in the “CDP Back to Basics” webinar series.
