Practical Tips for Data-Driven Working

In his daily work as Data Strategy Leader, Louis de Roo has the privilege of getting a look inside a great many different companies and organizations. This means he learns something new every day about what does and doesn't work when putting data-driven working into practice. That also means the list below can't be complete by any stretch. But 'start small' happens to be exactly one of the tips he can offer. So this list will be regularly expanded in the future. If you're missing something, don't hesitate to leave a comment — Louis will try to address it in a future addition!

Start small

When designing and conceptualizing your first data product, the temptation is great to pack all your wishes and creativity into your first attempt. But that's exactly where many companies get stuck: with sky-high ambitions and expectations for that first dashboard, it's nearly impossible to set it up successfully, while as an organization you're still very much learning which processes, systems, and tools you need to bring together to arrive at a dashboard. So start with a small scope, giving yourself the time to get to know the process. This way, you avoid having to win a championship without any training!

Learn together

It's a cliche, but with data-driven working, the journey is just as important as the destination. Not only for the buy-in of your solution, but also for learning to work with data within the organization, it's important to involve all stakeholders from the start in reaching the goal. This way, you prevent working with data from becoming an exclusive party for the data department, and processes and responsibilities can grow along with the organization's data ambitions.

Who it's for

Looking at the same data is not the same as looking at the same dashboard. Where management needs an overview, that can feel too far removed or impractical for employees in an operational process. So ask yourself who's actually served by a dashboard 'for everyone' — there's nothing wrong with everyone having their own dashboard, as long as the source data is the same.

Precision

We often display data in detail 'because we can' — but in most fields, we don't base our decisions on differences past the decimal point. So think about what level of detail is relevant to the decision you're facilitating — and leave out the rest. This prevents discussions about rounding differences and hunts for the right figures.

What's true

As soon as two systems display the same measurement, a difference will arise. Even the tightest system integration will produce a rounding difference somewhere. And while most organizations can easily agree on which measurement they should use for a particular KPI, the big point of contention is often what the definition of that measurement should be. By clearly documenting from the start which data from which source, with which processing, you use, you not only prevent discussion, but also lay an important foundation for good data governance!

Quality

The most commonly heard reason not to work with data is quality: "there are errors in this dataset" is often used as grounds to disqualify the data, and therefore the data solution. The question is to what extent this is justified: every dataset contains errors, sometimes from operational processes, sometimes from technical integration, sometimes from manual entry. And for some applications of data, those errors are unacceptable, for example for a sworn auditor's statement or for chemical calculations. For most applications, however, a single small error in the data doesn't matter and doesn't change the picture at all. Don't confuse "as good as needed" with perfection!

Written by Louis de Roo
Data Strategy Leader