<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=701837490009717&amp;ev=PageView&amp;noscript=1">
Artikel teilen:

Visualization Tools do not Solve Data Problems. The Truth is: They are not Even Half the Battle.


"We want to do something with our data now, too." – "Very good! What are you working with?" – "With [please insert: QlikView / Power BI / Tableau]." – "No, I mean, how do you solve your data problems, like the integration of data from different sources, the (real-time) update process, modeling ..." – "That’s what I mean: With [please insert: QlikView / Power BI / Tableau]." – "Ah, hmm. [Embarrassing silence.]... and what else? I mean, underneath?" – "What?" – "Alright. Let's talk again in three months."


The problem with visualization tools like the ones mentioned above is not that they are bad or useless or without meaningful use cases. On the contrary: Many tools from that category are very strong in their field. The problem lies elsewhere: namely in the fact that large parts of the market still seem to be thinking, "To make something out of their data" means, in the first place, to produce even more colorful pie charts than are already fluttering through the office anyway. A fatal mistake.


1) Infrastructure first! Start with the work, then the colorful dashboards

Yes, visualization tools are great at visualizing (that's why they're called that). However, to fully leverage this expertize, they depend on a strong data infrastructure that they do not (or only partially) bring with them. Ideally, you need an already integrated, structured, high-performance database as a supplier. And to achieve that is anything but trivial because to get to this result you already have to have gone through all the infrastructural steps:

  • Building of high-performance hosting
  • Interfaces to source systems
  • Data integration and transformation
  • Data Modeling
  • Enabling queries that are as flexible as they are high-performing

That is the real challenge when it comes to real value creation from data.


Yes, of course, I can also query data ad hoc from various source systems using standard connectors in the said visualization and analysis tools and also possibly model data ad hoc, but this cannot replace a real data infrastructure for many reasons, at least in the medium term – the most important certainly being performance, sustainably targeted modeling and lack of breadth in usability (see section 2 below). A well-known dilemma results from this – based on the fact that as the owner of such a visualization tool I basically have two options:

  1. I allow each user to upload files or to connect systems on their own, and thus land directly in the not to be underestimated problem that there are at least as many data truths as source systems (or files).
  2. I only give my admins access to the connectors and throw the oh-so-valued advantage of flexibility and self-service into the wind – and there it is again, the usage bottleneck called BI administrator.

So, to use visualization tools like Qlik, Tableau, or Power BI to actually leverage their strengths, I need a solid data infrastructure. My actual challenges with data utilization will not be completely solved by these tools.


2) Creating value from data is much more than visualization

The second important aspect is the fact that analytical work and the preparation of fancy reports make up only a fraction of what a real data-driven work culture or real added value from data actually means. Yes, reporting is important. Very important even. (Whether it is so important that the reports must be in corporate colors and can be printed on DIN A4 paper at the touch of a button, on the contrary, is certainly controversial – and the opinions in old-established large corporations and younger companies are usually different.) And yes, real analytical work by analytically trained professionals is (at least at some stage) equally important. But with these two components, the largest part of the cake is not even distributed yet – again consisting of two aspects:

  1. Actionable insights based on operational data – transmitted through specialized operational tools that allow the corresponding users direct access according to their (usually not very strong) analytical skills.
  2. Intelligent, automated processes triggered by operational systems such as Email Marketing or Recommendation Engines, fed with well-structured data from the enterprise-wide, holistic data infrastructure. 

Only with this crucially important component: a data-driven working culture across the entire spectrum of operations, can true value creation from data succeed – and for this purpose, too, pure visualization and analysis tools are hardly or not at all suitable.


The moral of the story: QlikView, Tableau, Power BI – all these tools are excellent solutions for their area of expertize, collectively referred to with the keyword visualization. But to actually use them effectively, for example, as specialist tools for the business analyst based on a BI solution that covers the infrastructural and operational part of the data challenges, one has to be clear about what they can and cannot do, what they are and what they are not. So as long as I do not have to hear: "We are finally tackling our data challenges and have purchased [please insert: QlikView / Power BI / Tableau] for that” anymore, then everything is alright.


For a holistic look at the question of what it takes to achieve real added value from data, I recommend the following paper:

The Commerce Intelligence Blueprint: 7 Success Factors for Your BI Project. 


Subscribe to my blog to stay up to date on inconvenient truths from the field of Business Intelligence and analytics! 


Artikel teilen:

Related Posts

Einen Kommentar verfassen

Stay up to date