In a recent Dynata survey commissioned by Redpoint about the 2020 holiday shopping season, 70 percent of consumers said that they planned to shop exclusively with brands that personally understand them. A personal understanding does not necessarily mean the brand knows the consumer by name, but that it recognizes a consumer’s preferences and behaviors across every channel, and across an unknown to known customer journey. Consumers are tired of being bombarded with irrelevant messages and offers; a brand that possesses a personal understanding is able to cut through the noise, providing the right offer at just the right time and helping guide the customer journey to its logical conclusion.
The difference between a brand coming across as an unwanted nuisance versus a trusted concierge is the function of advanced identity resolution, a key feature of a customer data platform (CDP) for any intended use case where it is important to meet a customer with relevant information. For a retailer, this could mean presenting a hyper-relevant offer or action at the optimal moment and channel, based on the entirety of a customer’s interactions and factoring in the customer’s opt-in preferences and GDPR/CCPA or other regulatory requirements. For a healthcare organization, the relevant information may be subject to HIPPA compliance disclosure, such as a medication/dosage notification where a proper identity match is vital.
As a recognition process, advanced identity resolution operates robustly in the context of whatever a business’s use cases and constraints may be, which may be very different for different organizations’ customers, for example in healthcare care management versus retail digital personalization. But not all CDPs treat identity resolution with the focus that it deserves, for various reasons. With an understanding of advanced identity resolution components, we will see why the feature is such a valuable part of a CDP.
Identity Graph
An identity graph is a collection of “signals” – attributes representing behavior, transactions, or information linked together with identifiers – that provides marketers with a real-time view into everything there is to know about a customer. For a marketer to trust that a piece of information on Jane Doe fits with the rest of the identity graph for this Jane Doe, however, a host of things must happen. Getting these various pieces of identity right is what sets advanced identity resolution apart from a run-of-the-mill match.
It is essential, for instance, to accurately match the various elements that make up an identity graph. Customers engage with a brand across multiple devices/channels/platforms. Jane Doe’s interactions likely cover multiple browsers, emails, devices. She may have more than one loyalty account, phone number, address, etc. For every interaction, transaction or behavior, information is fed back to help piece together/update the identity graph. What device was used? Which browser? The identity graph ties together those fragments with the activity and/or action that was taken: an email sent, a social post, a website visit, an in-store purchase with a credit card.
Deterministic + Probabilistic Matching
To know which attributes are important for the purposes of identifying Jane Doe, the system must have both deterministic and probabilistic rules in place to handle all kinds of matching challenges such as nicknames, typing errors and identity relationships.
Deterministic rules are generally straightforward – trusting the consistency of a machine ID, for example, to recognize that all browser activity in a single web session is from the same device.
Without probabilistic matching, however, brands are still flying blind. If John Doe uses his wife Jane’s laptop to browse fishing gear, a brand that relies solely on a deterministic match may presume that Jane is in the market for a tackle box. Probabilistic rules can account for nicknames, partial addresses, other non-unique identifiers and human error, such as incorrect data entry (a misspelled name, the wrong address entered on a form, etc.). Or, as the fishing example indicates, householding. Probabilistic matching is required to sort out relationships, which a device – and usually a person – are not explicitly exposing. If a brand is going to infer behavior – “Jane likes to fish!” – it runs the risk of introducing friction into Jane’s customer journey if she’s suddenly bombarded with irrelevant offers.
Identity in the Proper Context
There are other ways the system should understand relationships in context beyond householding. One example is a B2B setting, where it may be necessary to know that Jane Doe works for Acme Concrete if the underlying use case is to send her information relevant to her role as CEO. IoT devices and sensors provide another example. A smart car, like a device, may be associated with a primary identity. A contextual understanding of relationships means that an accurate, up-to-date identity graph will likely be in a constant state of flux; Jane may go work for Acme’s competitor, she may buy a new car, she and John may divorce, etc.
Advanced identity resolution not only makes calculations about what is (or is not) a match, it also should be robust enough to handle fragments, or pieces, that arrive in real-time and determine an appropriate action, such as breaking up a prior match, discarding data that you can’t keep or don’t want (to satisfy a regulatory requirement, for instance) or performing an aggregation.
For example, if you’re tracking details of a web browsing session for the purposes of making calculations about affinity, intent, total viewing time, etc., perhaps the calculation itself – essentially an aggregation of the visit – is what you keep versus the detailed log of the visit. An understanding, then, of how you’re going to use fragments that make up pieces of an identity is another function of advanced identity resolution that, like changing life events, is also fluid in nature.
Basic Identity Resolution? The Customer Notices the Difference
Calling it “advanced identity resolution” – which is what is offered in the Redpoint CDP – implies of course that there is a more “basic” identity resolution. While true, one can argue that merely matching known identifiers without probabilistic matching, householding, or other capabilities discussed here (accounting for human error, multiple devices, accounts, etc.) makes a “basic” edition hardly worth the name.
The reality is, today’s always-on, connected consumer has little patience for a brand that cannot follow and recognize them across channels. A consumer expects a brand will know they have multiple devices, ID’s, or accounts, or that they’ve changed jobs, have moved or just had a baby. To the consumer, it’s incumbent on the brand to figure it all out. That’s advanced identity resolution. Basic just doesn’t cut it, not when the goal is to really understand a person in order to deliver accurate, relevant and timely actions, offers or information that respect the customer’s preferences, opt-ins and membership in a certain group (a business, a household, a loyalty club, etc.).
Advanced identity resolution puts the “customer” in your Customer Data Platform.