The Marketer’s Guide to Identity Resolution: Is Your Data Hiding Things from You?
As a marketer, you know that the pinnacle of personalization is your ability to map your customer’s omnichannel journey to create their custom buying experience. To do this, you need full visibility into the fact that the customer who’s standing in your store now comparing shampoos on her Samsung Galaxy is the same person who opened your special offer email with that 50-percent-off coupon. These days, personalization is key to your strategy, and the last thing you want to do is send your customer a promotion for a product she won’t like—or worse—already owns.
Learn more about putting customer identity resolution to work in your organization
Unfortunately, companies are stuck with customer identities that are trapped in channel-specific silos—for example, web marketing systems that identify visitors with cookies or email systems that use a a single notion of identity: the email address. Of course, a person can have multiple email addresses, multiple mobile identities (by device), social identities, and more. But the problem is, all that data is usually living in distributed systems that don’t often talk to each other.
The REAL Martech Need: Complete Profiles, Constantly Updated, for Everyone
What you need is an identity resolution solution: a reliable way to recognize Android-9r7f9e8j as JaneSmith@gmail.com who opened your email this morning, and was on your website last week. The demand for identity resolution solutions is increasing in an age where consumers interact with brands on multiple devices in multiple channels, sometimes many times over the course of their day. Google research shows that eighty-two percent of shoppers consult their phones about purchases before they make them in a store. And according to DigitasLBI, the typical customer uses four different devices within the same purchasing cycle.
What is Identity Resolution?
To clarify, identity resolution is the data management process of analyzing disparate data sets to resolve a customer’s identity through different attributes, such as email addresses, device IDs, cookies and more. These attributes identify people with both personal and anonymous information and the process of stitching them together is typically automated through software solutions. The resulting data graph, assembled through algorithmic and statistical analysis, creates a persistent identifier that is used across systems. This identifier is then used to map new data about each individual into a comprehensive profile.
Companies with an effective customer identity resolution solution understand the unique ways a customer engages, and can unify those engagements into a cohesive customer profile.
Another important part of identity resolution is to ensure that this step is performed in compliance with consumer privacy data protection rules, which might include the GDPR or CCPA. The better identity resolution solutions have built in these capabilities or partner with other firms to provide these checks and protections as part of their complete packages.
Why Do Businesses Need Customer Identity Resolution?
Before identity resolution algorithms are applied, disparate data collected on a person across marketing silos usually fails to link back to the same individual profile. This makes it more likely that a person will receive different offers and messages on different devices—reflecting poorly on your company. Understanding that multiple customer records are the same person can also help organizations understand consumer behavior better, resulting in more personalized and relevant interactions. This is the unified customer profile or the ‘golden record,’ which has helped companies like Amazon deliver highly personalized experiences.
How to Use Identity Resolution in Lead Generation Management
Most importantly, identity resolution helps an organization narrow down its lists of customers and prospects that it markets to—saving time and recouping dollars spent. Many marketing organizations go without this valuable golden record and end up doubling their advertising and email (ESP) spend very quickly.
CMOs are well aware of the need for identity resolution. According to Winterberry Group, 40 percent of brands said that better identity recognition capabilities for matching multichannel consumers would do the most to advance their organization’s omnichannel marketing efforts.
How Does Identity Stitching Work? Probabilistic vs. Deterministic ID Matching
In the process of stitching together profiles, marketers use two types of matching algorithms: deterministic and probabilistic.
Deterministic ID Matching
With deterministic matching, customer records are matched by searching for equality across identifiers such as hashed email, phone number, or logged-in username. This high-confidence approach works best when first-party data is readily available.
Probabilistic ID Matching
With probabilistic matching, profiles are matched through an estimate of the statistical likelihood that two identities are the same customer. The identifiers could be such things as IP address, device type, browser or OS. Probabilistic matching can be less certain, and stakeholders must decide the level of confidence necessary to determine a positive match. However, it can be useful when first-party data is limited, or when reach is a priority.
To maximize results, Treasure Data’s enterprise CDP supports both approaches to identity resolution, depending on individual marketer’s needs. Our sophisticated algorithms deliver accurate identity unification at scale, while minimizing the number of direct comparisons needed. This approach reduces the compute and cost necessary to resolve disparate customer identities.
Identity Resolution Empowers Predictive Modeling
An additional benefit to identity stitching is the ability to do more accurate predictive modeling, producing the “training data” necessary to identify “lookalikes” within other customer sets. With automated predictive modeling built into an enterprise CDP, the model-building engine correlates hundreds of profile attributes to provide a recommended list of the most meaningful profile features. Marketers are free to adjust the model by adding more attributes to include, or by deleting suggested attributes. Of course, to build a reliable predictive model, you first need a large set of known customers to use as training data—which is why identity resolution is a key component.
Unified Customer Profiles Drive Efficiency Across the Martech Stack
Unifying your customer profiles to create golden records helps your entire martech stack work more efficiently. With unified customer data all in one place, you can truly personalize and customize every marketing interaction. Instead of mass emails, you can tailor your marketing based on your customers’ preferences or most recent activities. If you know their recent searches, for example, you can send out a campaign with special offers on relevant products. Personalized campaigns will boost both customer loyalty and revenue. And you’ll avoid embarrassing gaffes and irrelevant offers.
In addition, all your reporting and analytics will become more valuable with ‘truer’ data. With every golden record, you will get to see each individual’s unique interests and behaviors, not a segment built from a set of data based on good guesswork. Unified profiles help you—at last—to get the right message to the right person, at the right time and place. If you’re ready to learn more about putting customer identity resolution to work in your organization, download our white paper, “Recognizing Your Customers As Individuals.”