What is a metrics taxonomy? Why is experimentation so hard? Pandas' evil twin, the divergence of analytics and ML architectures, and taking back the discourse on the business value of data.
The connective thread that emerges for me through each of these different areas of discourse from self-service taxonomies, to experimentation, to underlying architectures, and finally to enduring (with humor) through change is... the future will benefit from a more ✨purple✨, holistic approach to what an organization wants to know about itself.
I find myself anchored to that question as I approach new work -- what do we want to know about ourselves? The answer to this question starts to define the data, systems, and process required as you and your team do and build things.
Separating the work of defining your analytical and information systems from the work of the activities that produce the data is a disservice to everyone imo (and experience).
Taxonomies are a great starting point, but the relationships between things is actually where ontologies become relevant, both informing the strategy of your information architecture. I argue that prioritizing this in the long-term is the the ultimate strategic advantage (the moat of the future if you will).
Take it in strides, prioritize the most business critical information, but START NOW, and make everyone at least aware of the work and the goal...in the future, every team is a data team, working from a shared ontology. This is the key to unlocking systems of intelligence, and I believe that the metrics and entity layer have the promise to accelerate this pursuit.
The connective thread that emerges for me through each of these different areas of discourse from self-service taxonomies, to experimentation, to underlying architectures, and finally to enduring (with humor) through change is... the future will benefit from a more ✨purple✨, holistic approach to what an organization wants to know about itself.
(https://blog.getdbt.com/we-the-purple-people/)
I find myself anchored to that question as I approach new work -- what do we want to know about ourselves? The answer to this question starts to define the data, systems, and process required as you and your team do and build things.
Separating the work of defining your analytical and information systems from the work of the activities that produce the data is a disservice to everyone imo (and experience).
Taxonomies are a great starting point, but the relationships between things is actually where ontologies become relevant, both informing the strategy of your information architecture. I argue that prioritizing this in the long-term is the the ultimate strategic advantage (the moat of the future if you will).
Take it in strides, prioritize the most business critical information, but START NOW, and make everyone at least aware of the work and the goal...in the future, every team is a data team, working from a shared ontology. This is the key to unlocking systems of intelligence, and I believe that the metrics and entity layer have the promise to accelerate this pursuit.
taxonomy vs ontology: https://www.earley.com/insights/what-difference-between-taxonomy-and-ontology-it-matter-complexity
systems of intelligence:
- https://twitter.com/jillzzy/status/1466062420795957249
- https://news.greylock.com/the-new-moats-53f61aeac2d9
- https://www.reforge.com/brief/why-systems-of-intelligence-are-the-new-defensible-moats
- https://www.protocol.com/enterprise/saas-systems-of-intelligence