I really liked Robert Van Ossenbruggen’s idea about top-down and bottom-up approaches to data and insights, captured in his visual below.
The concept defines a subtle but fundamental difference between bottom-up ‘data-driven decision-making’ and top-down ‘decision-driven analytics’ (for which Robert credits the book of the same name by Bart De Langhe and Stefano Puntoni). The former starts with the question ‘what data do we have?’. The task is then to decide how best to structure and organise the data, understand the patterns and what they mean, which can in turn inform your decision/s.
The top down approach starts from the business question. Understanding what decisions you need to make in order to deliver your strategy (which is, after all, a series of choices). Once the decision choices are clear, you can then look at the hypotheses, tests, information and data that you need to support these decisions. As Robert says, this helps to ensure that a team thinks more deliberately about what data they need, helps avoid simply using the data that is most conveniently to hand, and reduces interesting but unhelpful distractions that may arise.
This pyramid representation reminded me very much of the DIKW pyramid, which is a model I began using over a decade ago. I usually use it in the context of broader data strategy to talk about the need for a good foundation of clean, usable, accessible, relevant data, and how raw data has little value until you structure, organise and categorise data to create information.
Whilst you may be able to answer simple who, what, where, when questions with information, much greater value comes when you apply meaning to that structured data to define and understand patterns, connections and relationships. This turns information into knowledge which can answer the ‘how’. But it’s only when insight and knowledge are made actionable that the true value of insights are realised and we reach wisdom, which answers the ‘why’ question.
Superimposing Robert’s approach onto the DIKW pyramid fits really well. If we imagine strategy sitting above the pyramid then a top down approach seeks to define what decisions or choices within that strategy will help us to achieve our objectives. Then we need to understand what insights will be needed to make good judgements, and what information and data are required to derive those insights. Where we don’t have relevant information to help inform us, we may want to create hypotheses that we can test in order to generate fresh information. Testing different alternatives helps avoid the perils of confirmation bias. As before however, the bottom-up approach starts with the data.
Both a top-down and a bottom-up approach are valid in different contexts but the starting point is very different. The former is more strategic and closely linked to the delivery of core objectives, the latter is perhaps more executional and more likely to be used to define insights and actions within a defined space. For this reason we might begin with a top-down approach to define the domain, and then use a bottom up approach to understand necessary actions within that domain. But the two are interrelated – over time we might find that a bottom-up approach can inform a new set of decisions required to deliver our strategy.
A version of this post was also published on my weekly Substack – To join our community of thousands of subscribers you can sign up to that here.
To get posts like this delivered straight to your inbox, drop your email into the box below.
Leave a Reply