ANALYTICS WITHOUT THE SILOS
One of the key challenges to data analytics is finally attaining the long-sought, 360-degree view of customers. Are we getting any closer to this vision? Industry leaders say it’s still a work in progress.
The challenge? “Most companies deal with data silos and disparate tech that make gaining a complete understanding of their data impossible,” said Minnick. “This not only presents a data consolidation challenge but opens up data quality and governance issues too.”
In this environment, “enterprises collect more data than ever with less and less understanding of what they have,” Minnick continued. “We’re seeing lakehouse architecture become a prevalent trend for enterprises because it can unify all data types and apply a single approach to discovery and governance.”
For large companies with decades of legacy technology, “a big challenge in capturing a 360-degree view of customers and business operations is data silos,” said Williams. “To unlock data, it’s important to think of it as an enterprise resource rather than a business unit asset. Trapping data in silos doesn’t enable you to deliver enterprise insights and get its full value. Changing mindsets around data and practices is important and will take time, particularly if you’re reducing data silos that were built over decades.”
Data fragmentation and the increasing complexity of data environments are vexing challenges. “Data can be fragmented across data lakes, operational data stores, and cloud data warehouses,” said Malakar. “Data often exists in silos across systems and departments, each tracking different facets of customer engagement and business metrics. Breaking these silos requires unified data access that allows all users to find, understand, and trust data scattered across the business. Centralizing critical metadata enables organizations to overcome these hurdles and develop a comprehensive, accurate view of the business and customer touchpoints.”
In many of today’s environments, “customer data is frequently fragmented, messy, misaligned, and immense,” Jacobson agreed. “A customer could be using a different email address in the rewards system versus what is in the transactional system; the customer might be on social media with yet another different email address, and when they came into the physical store, they did not have any matching email address. Which email is the right one? What does the customer prefer you to use? These examples are at the shallow end of the pool.”
Data silos, data quality issues, and privacy concerns continue to hamper analytics efforts. “Data integration across multiple systems and formats—structured and unstructured—complicates capturing a holistic view of customers,” Bellur concurred. “LLMs can be a solution for unifying data, as their ability to process and interpret natural language data can bridge some gaps by automating data extraction, categorization, and sentiment analysis.
However, data governance remains a challenge due to privacy and compliance standards that require strict data handling practices.”
Addressing these issues “requires not only technological but also ethical frameworks to ensure transparent and responsible data use, particularly when handling personal and sensitive information,” Bellur noted.