Data-First vs Data-Driven
Two different ways of thinking.
Background
Of late, I’ve been pondering about the terms “Data-Driven” and “Data-First”, and wondering if there was a fundamental difference in meaning or simply semantics. As organisations lean into their Digital & Data Transformation journey, I have seen them latch onto these terms in their northstar vision statements and strategic intent.
I believe there is enough difference between the two terms, and so I dedicate my 60th article to unpacking what it means to be data-driven versus data-first.
(I write a weekly series of articles where I call out bad thinking and bad practices in data analytics / data science which you can find here.)
Putting Data First
When an organisation uses the term “mobile-first”, we get what they are saying. It means that their access touchpoints and applications are all built for mobile experiences first, and then translated or extended into other devices subsequently. Same thing when someone says that their organisation’s strategy is to be “digital-first” — it means that all their workflows will be designed as digital and straight-through with as little analog components as possible. So what does it mean for an organisation to say that they want to be “data-first”?
A quick scan of the internet reveals so much nonsense when it comes to trying to unpack this “data-first” term. I’ve seen one definition which says that “data-first” means putting innovation first and making sure your organisation is API-enabled to transmit / share data easily. Another rubbish explanation says that “data-first” means organising all your functions and processes around data. I continue to marvel at the bullshit that vendors try to sling, hoping to hit some money-laden C-suites along the way!
The terms “mobile-first” or “digital-first” are meant to embody design & built principles. We can anchor on a similar design & built principle for “data-first”. That principle is relevance and quality. Consider the example of building an application. You would need to (a) instrument for the data that informs you that the solution is operating as intended (i.e. relevance), and (b) ensuring that any data that is being generated through the application has to be error-free (i.e. quality).
Similarly, if you are putting together a new function within the organisation, then from a relevance perspective, you would need to define/instrument the underlying data to measure successful operation of the function, define/instrument the data input into the function and the data output (handoff) from the function. From a quality perspective, you would need to ensure that the function produces error-free output data (because that’s what you can control). In summary, applications, processes, and functions are described by their data requirements.
Driving with Data
On the other hand, the term “data-driven” evokes a process-oriented perspective. It’s premised on an organising principle, rather than a design & built principle. Processes imply a cultural mindset, and changing processes require a cultural shift. To be “data-driven”, organisations must ensure that their decisioning processes incorporate replicable data to achieve consistency, and are constantly evaluated to support continuous improvements. “Data-driven” is a way of working.
The challenge for organisations to be “data-driven” is the documentation of the decision processes — what data was used as input, how was it interpreted, what outcome did it generate, etc. There is an erroneous belief that some organisations have — that if they invested significantly in data platforms and AI tools that they could label themselves as “data-driven”. Without culture change, all those tools don’t and won’t matter.
An interesting point to note is that being “data-driven” isn’t about making sure you have data, but rather, making sure you use the best data you have at hand to influence and shape your decisioning. Being “data-driven” is about having an extractive mindset.
Conclusion
Being “data-first” isn’t the same thing as wanting to be “data-driven”. However, in many ways, the two things are very complementary. One is about creating a data-infused environment (availability of good data) while the other is about generating value from it (quality of decisioning).
Should an organisation pursue both? Yes, for sure. But to turn both into meaningful capabilities, you would still need to acquire a whole bunch of enablers. You can acquire tools, you can acquire storage, and you can acquire specialised talent. But in the end, the single greatest challenge would be enabling Data Fluency as the underlying enabler. Both “data-first” and “data-driven” require a whole-of-enterprise participation to truly matter, and people are notoriously difficult to mould when it comes to being more “data-enabled”.
