Building an identity graph helps brands generate a consistent, persistent 360-degree view of individuals and their relationships with those brands. But there are several critical steps to take to ensure success.
I was recently asked to define identity resolution, which triggered flashbacks to every time a family member or friend asked what I do over the last 20 years. The response to both means coming up with a simple, understandable description for a complex problem – to join data together brands need a common key.
Identity resolution is the ability to optimize and resolve all data points (online and offline) to create an identity graph with a consistent and persistent 360-degree view of individuals and their relationships with brands. The information this graph presents powers and improves use cases across the enterprise. There are five important factors when considering a first-party identity graph to power business.
- Build the identity graph at the enterprise level
- Understand how to improve identity resolution
- Make sure solution is configurable
- Ensure it is compliant with “privacy by design”
- It must be measurable
Building an enterprise-level identity graph is the most important investment a brand will make this decade. Why? A consistent and persistent definition of the customer is required to deliver exceptional experiences. It improves customer engagements (call centers, in-store sales), analytics, operations, and marketing – making it one of the brand’s most valuable assets.
Every platform across the enterprise has some type of identity process and creates identifiers necessary based on its particular function. To achieve consistency of how a brand engages with people across each platform, access to an enterprise identity graph fueling these platforms is key. An enterprise identity graph supercharges customer data platforms (CDPs), master data management systems (MDMs) and data management platforms (DMPs), putting the brand in control rather than relying on a thirdd party.
Improve Identity Resolution
If connecting the data sources across realms (offline/online), systems and platforms were easy there wouldn’t be a need for robust identity resolution engines. There is not an “easy button.” To connect data, there has to be a common piece of information to bring records together to determine if they represent the same person or if a relationship even exists. Many things can be done to improve matching.
- Improve the data quality and start as close to the time of collection as possible. Put processes in place before resolution to further cleanse, standardize, and fill gaps to optimize the data collected.
- Connecting online and offline data in a privacy-compliant manner requires the de-identification of the offline (known) data. Then the stitching may occur, utilizing de-identified first-party attributes and third-party referential identifiers.
- A brand should consider third-party sources of data to further improve the identity graph. This creates a strong spine within the graph that may be used to improve knowledge of customers and power prospecting capabilities.
Notice that I say “connecting” and not “merging” or “de-duping.” The purpose of an identity graph is to connect or stitch all the touchpoints and signals to a person. It is like a master key ring for each person with all the keys (identifiers) related to them hanging on. There may be multiple keys that are relevant at the same time, depending on the use case.
Experience has shown there is no “one size fits all” for identity. Each brand has its first-party data collected with unique keys, business rules, requirements, and use cases. However, there is a commonality within industry and country that provides a starting point.
A sophisticated identity resolution framework (engine) should be configured to support a brand’s specific requirements and be future-proofed. Note that I use the term “configurable” and not “customizable,” because that often means it can be done but that it will require development.
Other terms related to identity resolution are “deterministic” and “probabilistic” matching. Again, these matching techniques come back to the term “configurability” so a brand can apply the right combination of both methodologies to achieve the level of precision required to support a business use case. It may also be that a brand has both use cases requiring a high level of precision and others that are more about reach and that can afford to be looser.
The criticality of compliance both in the build and management of the identity graph itself AND in its enablement of downstream use cases cannot be stressed enough. From an identity perspective, the identity graph must at a minimum provide visibility to the source data, how and where it was collected, corresponding consent, and the ability to forget when requested. Poor compliance can result not only in bad customer experiences but is also a huge risk that may have monetary consequences.
The ability to measure the quality and value of an identity graph over time is a validation of the investment. It’s eye-opening to see the initial results of improved data quality and the opportunity to bring records together, identifying duplicates. The true value though is in getting it consistently right. It is in being able to improve the customers’ experiences because a brand recognizes them at every interaction and can show a relevant next best offer and improve call center support.
Identity resolution is extremely powerful. Accurate identity improves customer experience and increases revenue. Inaccurate identity often results in negative customer experiences, loss of business and compliance risks. There is not a single solution that fits all. If you are considering building a brand identity graph, engage experienced resources to assist you in evaluation. There are many options, and the one that is right for you depends on your specific use cases and requirements.