We are living in the Dataverse. Which means oftentimes, people associate the power of data with how much they have (ahem, big data). To quote Masayoshi Son, “Those who rule data will rule the entire world.” But more important than the amount of data in your repository, is the ability of your organization to effectively leverage that data to improve decision-making
If you were to ask, most leaders desire a company replete with capable employees who habitually interpret, communicate and take action with data when tackling problems (I mean, who doesn’t?). In short, they want a data literate organization. Ordinary individuals need real-time access to data in order to seize opportunities and maximize competitive advantages without barriers.
But the reality for many organizations is that when a business question arises, the most well-intentioned stakeholder might search for the answer in their go-to BI app, but more commonly, they come to an analyst to do it for them. Why? Because they don’t have access to the data — or worse — they have access and don’t know how to use it, thereby undermining data democracy.
Data democratization means providing hassle-free access to data in an easily consumable way. It also means that the IT and BI teams aren’t gatekeepers, nor are they the single source or even the starting point for all data analysis. To accomplish this, organizations need to circumspectly liberate or democratize data they previously kept confidential or exclusive to the analytics team.
So how can organizations both easily and safely become a data democratization power house? The answer: a foundational data model (FDM).
What is an FDM, and how will it help me achieve data democratization?
It’s one data model that properly combines and associates data from various source systems used across numerous business functions. It incorporates all the complex business logic, aggregations, critical data, calculated fields, application expressions and flags that are needed to tackle most business analysis, and produce any new application. It doesn’t have to be a BI app – though that makes it easier – you can use a database, a data warehouse, a set of application-ready QVDs, or any tool that accomplishes the same end-result.
Why should you implement a foundational data model within your BI architecture? Here are few reasons:
1. It creates a single source of truth
When you do not have the same consistent back-end behind all of your applications, you inevitably start each new BI project or request from scratch — reinventing the wheel with each deployment. This generates a large amount of wasted development and analytical resources, but more importantly, it creates distrust and confusion among your stakeholders. On the bright side, creating an FDM removes the burden of app development and complicated ETL work from your end users and dramatically increases data literacy as business users have just one data model to learn across every BI application.
2. Improved Governance
Reusing the same data model for every application takes the guesswork and duplicative effort out of ensuring the appropriate fields are being used. Business logic has been accurately reflected in calculated fields where data sources have been combined correctly and field naming conventions have been followed — data lineage can be viewed and so can any other governance activities. Ensuring data integrity and performing oversight become a breeze for your analytics team when data is centrally located instead of being spread across several conflicting dashboards. It's data democracy without the anarchy.
3. Cross functional alignment
Perhaps the greatest benefit to producing a standardized data model across your organization is that leaders and decision makers across functions look at the same dataset, labeled in the same way, derived from sources they trust and definitions they co-created. Instead of going into meetings armed with secret data or fighting about whether a chart or figure can be trusted, the refrain “I built this on the FDM,” settles all doubts.
Before the launch of the FDM at Qlik, we were wasting development resources with each new project, creating multiple sources of the truth, and falling short of our data democratization attempts. You might be doing the same — and it’s time to change that.
Questions about how to implement an FDM in your organization? Already done something similar and reaping the benefits? Let me know in the comments section below.
Long live the data democracy!