The term “Golden Record,” as a descriptor for a single source of truth for a customer profile, implies a certain infallibility, which is not to be confused with being immutable. Meaning that while a Golden Record will contain certain characteristics – namely identity and transactional components that, together, are a perfect representation of a customer for any given moment of a customer journey – those characteristics are fluid. Not only does the type of data collected differ for each business or industry, but the data are also constantly changing.
The power of a Golden Record as the foundation and launching point for a hyper-relevant, personalized customer experience is that even with a continual influx of new data, it is always in the context and cadence of an individual customer journey. With a Golden Record, marketers can be confident that minute by minute, second by second, they will always have the most recent and complete picture of a customer at every stage of a customer journey.
A Unified “Profile” vs. a True Golden Record
Marketers need to be wary of vendors who often overpromise and underdeliver on providing a single customer view, or another term they may use in place of Golden Record to mean a unified customer profile. What many really offer may be more accurately called “single view light,” which often stops at integrating customer data from multiple sources, and letting marketers decide what they need.
Many vendors ingest data from multiple sources and present it as a single view, but they require the data to have relevant keys to compile this “integrated view”. There is an assumption, in other words, that data quality processes are already complete and that it is ready to be made available across the enterprise.
There are several reasons why this approach is inadequate to keep up with a dynamic customer journey. First, customer data integration is not a synonym for real time. Integrating data from sources where the data being compiled are out-of-date (meaning, in some cases, seconds or milliseconds old) compromises accuracy. A complete profile becomes incomplete the instant data latency enters the picture, for the simple reason that a marketer will be unable to trust that the profile is an accurate representation of the customer. Because customer interactions with brands occur during dynamic, omnichannel customer journeys across both physical and digital channels, seconds matter. A profile is only as accurate – and thus trustworthy – as the recency of the oldest data.
Second, left to their own devices, marketers tend to default to the use of every available piece of data. The problem here is that data veracity and data quality are key components of a Golden Record. When data is cleansed, matched and merged at point of ingest (i.e., in-line), marketers are assured that the resulting profile accurately represents the customer they’re intending to interact with. A marketer, for example, should not have to deduce whether a “John R. Doe” is the same record as a “Jon Robert Doe” or if “123 Main St.” is in fact the same physical address as “123 Main Street”.
Depending on the business, or the purpose for the match, there could be a combination of deterministic and probabilistic matching at play, and often both are important to ensure a marketer or business user is communicating with the right person, household or business. Assigning persistent keys help ensure consistency. A retailer, for example, may determine with a high degree of confidence that a device used to browse a brand’s website belongs to one individual, but the credit card and shipping address used during a subsequent transaction belong to another individual in the household. Through a combination of deterministic and probabilistic matching, the retailer ensures a seamless customer experience for the right individual.
Another reason why data integration alone is not the same as creating an unassailable Golden Record is that combining data sources is not the same as making the resulting unified customer profile accessible to any person or system that needs it – at the moment it is needed. In addition to being in real time, with advanced identity resolution and data quality steps taken in-line, to be considered a true Golden Record a unified profile must be made available in real time to enable a marketer to engage in the cadence of a customer journey.
Constructing a Golden Record
A Golden Record in one industry may look vastly different from one in another industry, even though both will contain identity and transactional facets. A retailer or another transaction-based business will very likely use 30 days spend, lifetime spend, average transaction value, customer lifetime value and other similar metrics in a Golden Record, whereas a subscription-based business may be concerned with only the length of the contract and the billing cycle.
An organization that sells high-end products or big-ticket items – luxury cars, say – will create a Golden Record that looks very different from a company selling perishables or fast-moving products like laundry detergent.
A Golden Record worthy of the name will take a broader view of transactions to also include how a customer engages with content. Did they click on a link on the website, for instance? How long did they hover? Which pages did they visit, for how long and in which order? Did they open an email? A careful analysis of a complete transactional record will help ensure that a marketer engages with the right type of content – on the right channel.
Identifiers, too, are not restricted to what we usually associate with a customer – a physical address, email, name, etc. Rather, an identifier will include anything possibly associated with a customer – all devices and IP addresses, household information, etc. As with the transactional trail, important identifiers may vary by industry.
A Golden Record for a bank, for instance, will include all account numbers associated with a customer. The importance of various identifiers will also vary. An ecommerce company may have little use for a physical address, which will be very important for a company shipping a physical product.
Beyond the Golden Record
In addition to a Golden Record, there are other tools a marketer needs to consistently deliver a hyper-personalized, omnichannel customer experience. Automated machine learning is a requirement for drawing the best insights from the golden record in real-time and delivering a personalized customer experience at scale, while a real-time decisioning engine is vital for calculating a next-best action for an individual customer at the moment of interaction.
But it starts with a Golden Record as the foundation that supports everything else. How it’s created will vary by industry and by company, but each will also be unique in that it not only represents a specific customer, but a specific customer at a precise moment in time. With cleansing, matching and merging done at the moment of data ingest, no data latency, consistent updates and instantly accessible, a Golden Record provides marketers with confidence that they will always engage a customer with consistent relevance throughout a dynamic, omnichannel customer journey.
