There's a lot of discussion around what the modern data stack should look like, whether to use data fabric or data mesh, and the necessity of data modeling today. In my humble opinion, these conversations often miss the mark. The real questions we should be asking about any data architecture, technology, and capability should focus on business value and efficiency.
Key Questions to Ask About Your Data Architecture
- How quickly can an end-to-end data-driven business use case be delivered?
- A simple business metric should take minutes to build and publish, while more complex analytics should take days or a few weeks at most.
- How well can the people, process, and technology support business questions?
- Itโs not just about the tech. Aspects like user experience (UX) and data storytelling are crucial for supporting business needs.
- Can it support explorative data analysis and experimentation without long onboarding cycles?
- The architecture should enable quick and efficient exploration and experimentation.
- Are people spending more than 20% of their time on non-business value tasks?
- If so, this indicates inefficiency. People should focus on tasks that add direct business value.
- How can the business trust the output and understand what is happening?
- Simplify and make transparent the data pipelines and transformations to ensure business trust and clarity.
- How does the business ask different levels/types of questions?
- The architecture should support both knowledge-based ("what does it mean") and decision-based ("what do I do now") questions, not just historical data queries.
- How quickly can ideas move from experimentation to production?
- This process should be swift and ideally automated, taking a very short time from idea to production.
- Can non-technical users easily build data products?
- The system should cater to tech-savvy business analysts and leaders, data analysts, and researchers, not just hardcore techies.
- How quickly can new analytics types, techniques, or frameworks be included to solve business problems?
- Flexibility to integrate new technologies is crucial. Your data warehouse might be great, but if the next valuable business problem requires a graph solution, you need to adapt quickly.
Simplify and Focus on Business Value
To summarize, keep it simple and focus on the business opportunity. Arrange your activities around providing value to the business within the maturity of the organization and current capabilities. Start with a simple, value-based roadmap, and build a bridge between the present and future, all while focusing on delivering business value.
Practical Advice
- Ship small, ship fast: Start with small, manageable projects that can quickly show value.
- Iterate and grow: Prove the value of your approach and secure further investment to build more complex capabilities.
- Avoid over-engineering: Donโt start with a Rolls-Royce when a simpler solution will do. Understand the process and take a business-first approach.
Embracing Agility and Rapid Value Delivery
The speed and agility with which you can deliver analytics with trust are paramount. This capability will change the organizational culture, fostering a business-first, value-driven mindset. It's about responding to and driving the business askโputting customers first, then requirements, processes, and people.
Conclusion
Focus on the core questions that tie your data architecture and technology back to business value. By doing so, you'll ensure that your data initiatives are not just technically sound but also strategically impactful. This approach will enable your organization to thrive, adapt, and continuously deliver value in an ever-changing market landscape.
Are you asking the right questions about your data architecture? How are you ensuring your data capabilities align with business value and efficiency? Let's discuss and share insights to drive our organizations forward.