Customer data analytics and the empowered organization
Data, whether in a data lake, enterprise data warehouse or an operational data store, is not useful. More accurately, data are not useful in and of themselves. Data only become useful when used, and in that use, context and perspective are added to the discrete facts that data represent.
Knowing discrete facts like “Frank,” “Topeka,” and “purchase on 12/1” have little impact on the larger business. Knowing that “Frank” from “Topeka” made his most recent qualifying “purchase” on “12/1” – now we have something to work with.
This brings us to the Information layer of the "data, information, knowledge and wisdom" (DIKW) pyramid. “Information” in the pyramid can best be described as where perspective begins to be added, allowing facts to be connected. This transformation from data to information is typically the result of users overlaying meaningful context, from straight forward data aggregation or from analytics producing higher-order measures.
Some examples would be:
- Applying business rules that differentiate purchases by type, size or product into qualifying and non-qualifying
- Adding a single purchase into an ongoing transaction history with relevant RFM measures,
- Aggregating the sales into a meaningful report,
- Correcting and completing a customer’s name or other details based on multiple data sources,
- Enhancing “Topeka” by adding the correct county, state  and matching to associated geo-details.
In the example above, the additional context and derived content added to all three data points is what begins to make them useful.
In the DIKW model, the Information layer is the domainof customer data analytics.  Analytics (whether descriptive, statistical, data completion or inferential) are how the disconnected facts of the Data layer are enhanced to provide business value. Using customer data analytics, business users can determine a given customer’s identity and preferences, understand their history, and determine their position in the customer journey. This makes customer data analytics a required step to create value from all the collected data.
The Power of Customer Data Analytics
In my last post, I discussed the shift from traditional data warehouses to operational data stores; this evolution means business users can more directly access their customer data. With greater accessibility, marketers and others can use cleansed data to meet their goals without as much dependence on the IT department and scarce data warehouse resources. This, in turn, decreases time-to-insight and increases flexibility throughout the enterprise.
Data lakes significantly ease the infrastructure and operational costs of collecting and storing big data.  Capturing data, however, is only the first step. Holding onto data, in and of itself, doesn’t provide value; using data to provide information and make decisions does. But the simple reality is that data lakes are too unwieldy and useful content is buried in the detail. Most business-user-friendly BI and analytic applications are unable to access the data lake, and if they can, they are overwhelmed by the volume and variety of data sloshing around it.
While businesses have increasingly become overwhelmed by the tsunami of data in the data lake, the rise of the empowered customer has heightened the need for effective customer communications. Customers expect to interact with brands through multiple channels, at different times of the day, for different reasons. They also expect consistency in these interactions irrespective of channel, content or latency. Because of this, using data to build a complete picture of customer interactions is vital for long-lasting customer relationships and staying competitive.
Customer Data Analytics and the DIKW Pyramid
The data storage and computation advances that I wrote about in my last post have had a positive impact on analytics technologies. Improvements in analytics now mean that usable information can be generated from raw data as rapidly as the customer interacts with online and in-person systems.
The problem is that this unprecedented speed puts pressure on those within an organization who have to manage the information. The question is no longer “Can I generate usable information from my data?” but rather “Am I able to make use of the information that’s generated?”
In this multitouch, omnichannel environment, successful customer engagement systems must have reliable and timely access to real-time or very-near-real-time information. Conceptually, this is not a change – the customer communications must be inside the customer interaction cycle. Any after-the-fact decisions about how to treat or communicate with customers are demonstrably worse than useless. Similar to how data are only useful once context is added, information (analytics) only becomes valuable when customer-facing systems can access it and use it to inform and shape customer actions.
Being able to add context quickly can potentially be a positive sign, but the information created from aggregating disconnected data must be both usable and accessible to other technologies. So, while we’re still overloaded by data, we now also run the risk of being overwhelmed by information generated from our data. This means we must be judicious in which data we keep, how we use them and, perhaps most importantly, how our data and insights are used to drive customer communications.
All in all, there is enormous value to be had in using customer data analytics to weld disparate data into useful information. The challenge, however, lies in ensuring that the created information is usable, actionable and accessible for all units that wish to leverage it.
That will be the challenge for organizations that evolve beyond collecting raw data: they must understand that having the technical ability to create information doesn’t necessarily mean they should do so. It’s only when information is thoughtfully created and made accessible that it truly shines.
 There is at least one Topeka in Zambia and four in the United States (in Mississippi, Kansas, Indiana, and Illinois) – who knew?
 Read data analytics broadly to include BI & reporting, traditional descriptive and analytic statistics, and contemporary techniques such as machine learning, artificial intelligence, and real-time decisioning.
 Big data here defined in the broader “I know it when I see it” because it has high volume, great variety, and/or high velocity.