In the world of data science, it’s quite easy to see projects as a series of checkbox exercises where the only way through the project is to ensure you’ve completed all the items on the pathway. Unfortunately, this leads to narrow perspectives in the arena of problem solving. The freedom to explore horizontally or discover possibilities is hindered by following the recipe created for previous projects – limiting the value attained by the solution.
Is it possible to undertake a project and ensure that you deliver the maximum value possible to the client?
Firstly, what defines value? Is value merely the most effective way to solve the problem immediately in front of you while minimising cost and time-to-value? Let’s take Blockbuster as an example. Having been in the movie rental business for a while, no one was better positioned to capitalise on the potential of movie streaming than Blockbuster. They did not leverage their expertise and capture the streaming industry with a subscription-based model. Similarly, private space transportation services companies are moving toward market dominance without the years of experience the government agencies have.
This reveals some important aspects of value:
- Value is defined by usefulness – solutions without problems are not worth pursuing
- Usefulness is transient – it changes as time progresses (you have little use for a diaper now but when you were a baby…)
- Usefulness is based on perspective – pigeons, while excellent chess players, have little need to borrow your calculator
Sure, value depends on the immediate need of the client. But further value can be unlocked by determining why those immediate needs have arisen. What causes the need? Is the data, that you’re looking at, revealing the problem space in its entirety? Or have historical and industry bias led the client to believe that these data have and always will identify a solution?
The scientific method would have you backtrack to the beginning. You would need to prove that the problem you are solving is indeed the root cause and not an effect of some larger problem space. Rather than beginning by going deeper into the data, casting the net wider and reviewing other aspects of an organisation’s data may reveal the treasure you’re actually looking for. Data interoperability is fundamental in the creation of value. It is also the first paragraph in the contextualisation of data. A context, which is very dynamic, would require this broader system context for anyone to identify and solve problems as they arise.
Organisations are not static things – they evolve on distinct timelines. Solving a problem in varying circumstances requires a degree of adaptability. Solutions should be less targetted or singular but have sufficient variability to solve secondary issues caused by the solution. As a basic example, increasing the availability of a vaccine only works if the logistics of storage, delivery and checking the expiration date work too – just sayin’.
Ultimately, this methodology starts with trying to answer a single question:
“What should we be measuring?”
Follow your curiosity,
The Afrobots @ Afrobotix.