Keeping Up With Data #110
5 minutes for 5 hours’ worth of reading
From my studies of general mathematics I know that objects are defined by their properties, not names. Semantics matters, and what’s better to clarify a meaning than a formula codifying a definition?
In many companies, people casually use terms like customer, product, transaction, sale, without common agreement on the details. How many times have we discussed who is a customer at various meetings? These conversations get even trickier when colleagues from different departments are involved and share their point of view.
With the growing adoption of data-driven decision making, we see more people requiring access to data. With that often comes the freedom (and duty) to create meaningful metrics (such as number of customers or monthly sales per store).
But with the freedom comes responsibility! The metrics we define will be used in various analyses, reports, dashboards, or documents. Often without sharing the definition. And thus every consumer will assume their own definition of e.g., a customer or monthly sales per store.
Data is quickly moving from financial realm with well-defined metrics (often by law). Other parts of business need to be described with similar rigour.
And we should do it fast, because defining outputs of machine learning models will be even trickier than the deterministic definitions of basic business objects.
Describing the way a business works, operates, and creates value goes a long way. It enables many data and analytics initiatives, but it also highlights opportunities to re-design business processes.
Both are incredibly valuable opportunities.
This week’s reading list looks at data strategy trends, metrics layers, and digital transformations.
- 7 enterprise data strategy trends: As the world of data and analytics is constantly evolving, companies must react and adapt their data strategies. The latest trends are bringing real-time data, growing demand for internal data access, taking a strategic approach to external data sharing, building cross-functional data teams, and more. My experience tells me its worth keeping an eye on the latest trends, but adopt them only after a careful consideration and at the right time. Every organisation is different, with different needs, and at a different stage of data maturity. (CIO)
- The Jungle of Metrics Layers and its Invisible Elephant: Metrics layer is a “layer where you would get to define standard metrics once, ensuring consistency of definitions, whether accessed using BI tools, queried from Jupyter notebooks or retrieved in other ways.” But how does it look like? Should it store data or just be a pass-through? How to define metrics and its semantics (how does a metric relate to other dimensions)? How should we query a metric layer? (aurimas)
- In the trenches of digital transformation: “Digital transformation is similar to replacing a propeller engine with a jet engine while flying at 40,000 Feet.” Its success depends on the CEO who needs to be the executive sponsor for the transformation, nobody else. What was pleasing to see was the use of data and analytics being named as one of the two biggest opportunities in digital transformation. Companies should aim to create strong foundations with value-creation applications on top (the mythical equilibrium between offensive and defensive elements of a data strategy). (Pollen Street Capital)
Enjoy the weekend!
