If you’re in the market for a data fabric solution, you should be looking for one that’s optimized for both online operational intelligence, as well as offline business analytics.
If you ask data pros, they’ll probably tell you that data fabric solutions are designed to enable business analytics and the meaningful insights and outcomes that derive from them. Many articles have been published about data fabrics supporting business intelligence, machine learning, predictive analytics, and trend analysis, all offline activities performed by data scientists.
Through all the hubs, the value of data fabric solutions for operational use cases, like credit scoring, churn prediction, customer 360, data privacy compliance, fraud detection, and real-time data governance. All depend on accurate, complete, and fresh data. For example, you can’t depend on a churn prediction algorithm based on last week’s data.
There’s no need to use one data fabric solution for operational intelligence, and another one for business analytics. The right data fabric should be applicable to both.
Online operational intelligence is super important
Hypothetically speaking, take the case of the “Los Angeles Medical Center” that, through its own analytics, has determined that the odds of a Beverly Hills resident, at a specified income level, moving to a different clinic are less than 1.3%. Say the medical center ran its offline prediction models at 1 am, but at 9 am its portal went down just as David, a Hollywood producer, and Beverly Hills homeowner, was about to order a pre-flight COVID test for his ailing wife. For the past half hour, he’s been unsuccessful in reaching anyone via the LAMC hotline, hanging up each time after a 7-minute wait. Obviously, by now he’s quite upset. By fully understanding David’s journey and behaviors – in the form of real-time inputs to the clinic’s churn model – up-to-the-minute operational intelligence would be super useful to the service agent about to take John’s call.
Offline business analytics, though essential, is not enough
With a good set of data analytics tools, marketing analysts can use a data warehouse to define a particular segment of customers for a specific advertising campaign. On the flip side, data scientists can use that same DWH for building a churn propensity model. In the medical center use case cited above, marketing would never normally target Beverly Hills homeowners with retention campaigns. So when someone like David calls customer service, he would not be tagged as a high-risk to churn. However, by treating David as an ordinary member of his market segment, under extraordinary circumstances, the LAMC could very likely lose a valuable customer.
The right data fabric solution serves both online and offline functions
The combination of 2 eyes in human beings, and many other mammals, allow for depth perception and peripheral vision. Both are critical functions when driving a car, for example, where split-second decision-making is essential.
At the same token, a data fabric that maximizes the depth of understanding, as well as the field of vision, for each and every customer, would not only provide enterprises with clean, fresh data for online operational intelligence (“depth of understanding”), but also for offline business analytics (“field of vision”).
On the one hand, serving a business entity’s real-time data into an analytical model is necessary for hyper-personalization, and its resultant insights and recommendations.
On the other hand, the data prepared and delivered for offline analytics would be complete, because it contains everything there is to know about a customer or any other business entity (such as a vendor, loan, order, credit card, or virtually anything else important to the business). In many cases, a customer’s data is fragmented and stored in disparate systems, making it extremely difficult to locate and collate, so a 360 customer view is essential.
The data fabric solution designed for both operations and analytics
The ultimate data fabric should provide complete and trusted data for operational use cases and business analytics. The following 4 steps illustrate how operational customer data is transformed into business insights in real-time.
- Data fabric delivers clean, fresh group data, based on a 360 view of business entities – e.g., customers, products, retail outlets, etc. – to data lakes or data warehouses. Accessing these big data stores, data analysts employ Business Intelligence (BI) tools for trend analysis, customer segmentation, and Root-Cause Analysis (RCA), while data scientists define Machine Learning (ML) models.
- The resultant ML model is deployed into the data fabric, as a web service that can be executed on demand for an individual business entity.
- Data fabric executes the ML model in real-time, upon request from an operational system, feeding it the individual business entity’s complete and up-to-date data.
- The ML model output is returned instantly to the requesting application and stored in the data fabric, as part of the business entity, for future analysis. Real-time recommendation engines can be invoked by the data fabric for the delivery of the next-best actions.