Identity Resolution Explained

December 5, 2018 — by MediaMath    

Customer data and identity resolution are complex. Clients I work with are often confused about which different types of data are available. They wonder which solution sets they should be exploring and on which capabilities they need to focus. So, let’s start with the basics. What types of customer data are out there?

  • Anonymous: Here you process data with the aim of irreversibly preventing an ID from being attached to the individual to whom the data belongs. Data becomes anonymized when there’s no way to identify the individual behind the data—even if you add more data.
  • Pseudonymous: In this case, you replace any identifying characteristics of data with a pseudonym. For example, a consumer might appear in the system as “AB1234.”
  • Personally identifiable: This is data that could potentially identify a specific individual, distinguishing one person from another. Either this data identifies the person, or you can identify them by combining this data with other data.

So, what does identity resolution look like in real-world examples?

Let’s say that I shop with Company X.  If I walk into the Company X store and pay cash, I do not provide any information. So, the company can’t identify me in any way. It will record my transaction, but it will be completely anonymous; the store has no way of understanding anything about me.

But imagine I browsed the Company X website on my laptop without logging in or purchasing anything. Then Company X would likely be able to identify me pseudonymously by the cookie trail that I leave as I browse through its website.

The brand can’t connect that browsing behavior to me personally. But it does have an identifier (a cookie). It also may have a set of identifiers (cookie, device ID, IP address, etc.). It can use those identifiers to connect behaviors, context and information about me pseudonymously.

Finally, let’s say I decide to buy something from the Company X website. I log in to its website using my email. I also decide to provide my name, home address and phone number, so that they can ship me the items I purchased. The brand now has a robust set of personally identifiable information on me (email, first name, last name, home address, phone number).

Welcome to the world of identity resolution. Here, marketers try to keep track of disparate signals about a consumer and resolve them as much as possible into a single identity. That’s how you build a single view of each consumer.

But, the aggregation, storage, access and uses for the separate types of information are different. That means the solutions for the data are also different. Anonymous data by nature is only useful for reporting. You can’t use it for targeting or marketing activation. You can use pseudonymous data for marketing measurement (connect all sales for, say, a cookie to marketing sent to that cookie). But, you can also use it for optimization (making smarter marketing decisions based on real-time customer behaviors) and activation (targeting on digital marketing channels using cookies, device ID, etc.).

You can use personally identifiable information in most of the same use cases as pseudonymous data. But regulations and laws limit what companies can do for measurement (connecting online marketing to offline sales) and activation without a consumer’s expressed approval

In the next post, we’ll look at the actual technical solutions that marketers use to aggregate, connect and profile customer data. We’ll examine how they can activate data for personalized, targeted marketing outreach, as well as how the worlds of pseudonymous and personally identifiable play separately and together.