As announced in our last week’s blog post, in this series we are aiming to share our learnings and experiences on how to successfully approach big data projects and continuously transform into a data-driven company. There are a few things to be aware of and focus on – for today it will be the strange but important animal “use case”.
We have observed that big data activities in companies often start with building up expensive and complex infrastructures to store and manage all the different data assets – the birth of a beautiful data lake. What is usually not in scope are the needs of important stakeholders once the data lake is built. In other words, very often there is no clear understanding of important use cases that should be addressed with the proposed data-driven solution. However, this understanding is crucial in order to make reasonable decisions in an area that is as complex and as dynamic as the area of big data analytics. Do I want to leverage data in order to be able to offer more suitable and relevant products and services to my clients? Do I want to fully automate certain marketing decisions? Or do I want to introduce predictive maintenance algorithms in my factory?
Fail fast, learn fast – what we can learn from ROBASO
No, ROBASO is neither a serious illness nor a new fancy name of a fully automated robot cleaning your dishes or bathroom. ROBASO is the abbreviation for an IT project initiated and recently stopped by the “Bundesagentur für Arbeit” (German federal agency for work). Don’t get us wrong – this is not about bashing federal institutions – there many similar examples from the private sector. The project’s objective was to design one common IT platform to harmonize 16 different standalone applications. An example of an inefficiency that ROBASO would address was the requirement to manually update each of these 16 standalone applications if a job seeker were to find employment. In general, ROBASO was a very good idea – however, after 6 years and 60 million Euros of investment the project was stopped.
When the new platform was finally tested at the beginning of this year, it was realized that the new software was far too complex and did not meet with the real requirements and daily workflows of its users, federal employees. For example, if the bank account of a customer changed, the system did not adapt automatically and the customer’s profile had to be set up again completely from scratch.
If we translate this into what is mentioned above, obviously there are three key elements missing:
This is why we are convinced that use case exploration and definition needs to be put into the center of all activities when working with data driven marketing solutions. Determining the feasibility of use cases depends strongly on available data, suitable processes and appropriate frameworks. The simultaneous evaluation of use cases, data requirements and data realities are of crucial importance in ensuring the success of a Big Data project. In our next blog post we will introduce the approach of data-driven use case exploration that has worked very well for us.
Do you have other examples of Big Data projects that failed due to missing use cases? We look forward to hearing from you!