Let’s See It All.


Anyone involved in organizational analytics and planning — we will call this person the Knowledge Worker — is inevitably concerned with data integrity. The knowledge worker may study the workforce through key performance indicators (the present state), or be involved in strategic planning (towards a future state). More likely, the knowledge worker is looking at how a multitude of data factors into specific organizational projects — re-orgs, mergers, spans and layers optimization, retention management, across divisions and geographies.

It can be a complex, complicated endeavor. Complex is acceptable, complicated is not.

In December I wrote about organizational “big data” and data theory. The conversion from raw data to useful information involves a process of structuring, ordering, arranging, and clarification. This transformation is done through a process of design. Design is what makes the world of the arbitrary a world of understanding and accessibility. Another part of this design process could be data reduction, or, selecting portions of data sets that are seen to be more contextually relevant to the inquiry at hand. That’s expected, but the stakes are much higher now: In selecting data, there’s a high risk of confirmation bias, evidence suppression, cherry picking, even the risk of making the data less valuable by pulling it out of context.

Edward Tufte, Yale professor emeritus of analytic design, has made a career of pressing for statistical honesty and integrity in the design of information. In a 2006 essay, “The Cognitive Style of PowerPoint: Pitching Out Corrupts Within,” he takes to task the most pervasive tool we use in corporate and government bureaucracies today: PowerPoint. His argument essentially states that the tool itself, made worse by its pre-packaged layouts, forces its users to reduce evidence, damaging data integrity. He argues that in technical reporting, PowerPoint can be most harmful, pointing to NASA, who used PowerPoint to present data on the ill-fated Columbia space shuttle before it returned to Earth. It’s a damning essay. But his point is this, and it’s one that he has made for years prior to Columbia: present all the evidence, in dense form, clearly structured, free as possible from biases (one could argue this is never possible), and allow the audience to learn for themselves and ultimately draw weighed conclusions.

How many org charts have we all seen presented in PowerPoint decks? Sure, many of these are in PowerPoint because they are being delivered to an audience around a projector. But so many others exist there because PowerPoint is often the default tool that business people turn to for drawing boxes and arrows. The problem: an insane level of data reduction and distortion. Add other data to the org chart slide and one is quickly forced to compress, re-arrange, and filter out important evidence.

The presentation of organizational structure and workforce data demands tools that are comprehensive, honest, and data-dense. They also need to be designed in such a way that enables user insight while avoiding overload. The complexity of large organizations is a perfectly acceptable thing; it just needs to be managed and presented in comprehensive ways so the knowledge worker can make informed decisions on his or her own.