Supercrunch Blog

Dr. Ralph Wirth May 8, 2017 Big Data
Big Data, Broken Promise? Our learnings on how to avoid Big Data frustration
Part 3: Data meets use case – how to analyze which use cases to focus on

In the latest blog post of our “Big Data, Broken Promise?” series, we have shared our number 1 observation of why numerous Big Data projects fail: They start without a clear understanding of the use case(s) that should be addressed and without sufficient interaction with the final end user or target group. Therefore, we suggest to begin every Big Data project – in particular if potentially large investments are involved – with a thorough use case exploration and evaluation.

But how does this use case evaluation work in practice, and how can you make sure that you are not coming up with use cases that you cannot address due to data restrictions?

Throughout our Marketing Analytics consulting projects, we have had very good experiences with Data Design Thinking workshops involving key stakeholders on the client side. Depending on the project objective, the group that attends this workshop may, for example, consist of future users of the to-be-developed data-driven solution, data “owners”, data scientists, consultants, and a professional moderator.

It is important that these workshops follow a structured approach, supported by proven templates, and should focus on the following four key tasks:

  • identifying use cases,
  • defining data requirements,
  • evaluating (possibly) available data sources, and
  • prioritizing the use cases based on the data evaluation.

 

Our Data Design Thinking approach is an adaption of the classical Design Thinking process:

  • In the first phase, we empathize with the target group of the planned data-driven solution – we want to deeply understand their daily work and in particular their challenges that might be addressed by a data-driven solution.
  • In the second ideation phase, initial ideas on how to address these challenges are collected.

 

Phases 1 and 2 should not be restricted to feasible scenarios. Therefore, they usually result in quite a number of potentially promising use cases.

  • It is in the third phase, the evaluation phase, in which the important question on whether certain use cases are feasible or not is addressed. When it comes to our focus area of developing data-driven solutions, one key dimension to be evaluated relates to required, suitable and available data for each use case.

 

It is certainly not a given that all available and potentially suitable data sources are known and understood. In these situations, structured data audits and data exploration activities are required. While the audits (identifying data sets) are often carried out collaboratively by data specialists within our clients’ company and SUPERCRUNCH’s data consultants, data exploration work (diving into the data assets to evaluate key metrics) is usually done by our Data Scientists.

  • Within phases 1-3, a prioritized list of use cases and possible solutions is developed, taking detailed insights on required and available data sources into account. Is it now time to develop a full-blown, fully automated solution for each of these use cases? Again, we follow the Design Thinking process and suggest a fourth phase, the testing & prototyping phase, which involves developing a simpler version of the envisioned data product, focusing on the most risky assumptions. The main purpose of this phase is clearly to reduce risk.

It is important to note that the phases, especially phases 1-3, are not strictly sequential. Discussions and outcomes of one phase may very well influence another process step. A skilled and experienced moderator will ensure the required guidance. There is also a broad set of templates for structuring Design Thinking workshops available (e.g., on http://www.stattys.com/). Below, you can find a simplified version of a template that we use for evaluating Marketing use cases in the light of available and suitable data.

We are looking forward to getting your feedback and to understanding how you prioritize your Big Data use cases!