• Data silos are more than a technical issue—they stem from business structures and company culture.
• Unlocking the most valuable silos first can drive meaningful business outcomes like revenue growth and cost reduction.
• A centralized, flexible data platform ensures consistent access, faster innovation and lower infrastructure costs.
• Shared data domains and a company-wide governance framework reduce redundancy and improve data quality.
• A well-maintained data catalog helps teams find, trust and use the right data, improving decision-making and compliance.

• Lost revenue opportunities: Consider Miral, which develops and manages Yas Island, an entertainment destination in Abu Dhabi. Miral’s plans to cross-sell and upsell attractions across the island were thwarted by data silos—one per attraction—which limited how well the company could understand and engage customers. 
• Unnecessary operational costs: An example is when multiple departments manage separate data warehouses to run similar reports or when the marketing department buys ads targeting customers who have already bought the product. 
• Redundant infrastructure costs: Unaware of existing data sets that meet their needs, teams spend money to stand up and manage new point solutions.  
• Inconsistent measurement and reporting: If the customer relationship management (CRM) and finance systems apply different treatment to returns, discounting, promotional offers and the like, managers cannot confidently answer questions like, “What were yesterday’s sales?”
• Inefficiency: A sure sign is when analysts need to open multiple dashboards for insights to address a challenge or pursue an opportunity.

Data silos are primarily a people problem, not an engineering problem

With the high costs of data silos, why are they still here? The primary barrier is not technical: system integrators already know how to safely unlock data silos using data transformation and access management techniques. The persistence of data silos has more to do with culture. When faced with a business problem, a department’s first instinct may be to implement a point solution that, understandably, focuses narrowly on the department’s needs. A data fiefdom arises. Consequently, other departments are more likely to create their own point solutions rather than searching for an existing, similar dataset.

What’s the most cost-effective way to unlock more value from data silos?

No two companies have exactly the same data environment, so there is no single right approach. That said, some principles are universal.

Large enterprises will never eradicate all data silos, and that’s okay. Start by unlocking the silos that are causing the biggest business problems. Good candidates for a first project include:

• Standardizing the key performance indicators (KPIs) appearing in reports
• Delivering more value to customers by personalizing experiences based on context
• Automating manual processes and reallocating human resources to optimize those processes
• Creating new products from existing data, such as dynamic audience segmentation to predict churn or propensity to convert, or fraud identification in finance and insurance 

Don’t overinvest to build the perfect data platform—it doesn’t exist. Instead, prioritize flexibility, designing a data platform that can scale and accept new data types from new sources. Just as no one could have predicted the impact of COVID-19 or AI in 2019, no one knows what the next decade will bring.

After the data platform is built, ask Procurement to notify IT of requests for new systems that include their own data infrastructure–common with email, ERP, CRM and web content management systems, for example. This gives IT the opportunity to assess whether it’s more feasible to integrate the new system with existing data infrastructure. 

The optimal data domain structure varies by company. Common domains include “customer,” “transaction” and “product.” Industry-specific domains might include “insurance policy” in healthcare and “inventory” in manufacturing. Some companies will find it helpful to establish separate domains for the business-to-business and business-to-consumer lines of business. Companies subject to country-specific privacy regulations may benefit from establishing geographical domains. 

Assign a data owner for each domain and create a Center of Excellence (CoE) to help owners manage their data as a product, focusing on quality, usability and value to the internal customer.

Identify executive sponsors who can align the efforts of different organizations to share data. Select a steward for each data domain to ensure departments know that the data exists and provide guidelines for its use, such as consent management and retention policies. Agree on KPIs for governance, including data usage, consumption, production and availability. Set up certification processes to confirm that data meets organizational requirements for quality, documentation and freshness. Establish SLAs so that internal teams have confidence in the reliability, accuracy and timeliness of the data they use.

Make internal teams aware of available data by building and publicizing a data catalog that includes:

• Data provenance from creation to the current state, including transformations
• How the data should and should not be used 
• Certifications, such as HIPAA compliance, adherence to organizational privacy policies and consumer consent/permissions
• Date of last refresh

Don’t overlook the refresh date. A marketing team that doesn’t realize the model for predicting customer-churn was last updated 30 days ago might spam customers with the same offer sent two weeks earlier, resulting in a poor customer experience and unnecessary costs.