Explaining analytics: Learning before failure


ExplainingAnalyticsWhen I am discussing my new Explaining Analytics course with analytic professionals, I frequently hear, “I have a co-worker that needs the class.” That comment is much more frequent than – “Hey I need that class.”  The truth of the matter is we all need to continue to improve – all of us. There is a useful technique where we can all learn from each other to give this continuous improvement and can be applied anytime we are preparing for a presentation.  This technique focuses on learning from one of life’s greatest teachers – failure.  This technique provides a twist where we can learn before we fail, not after.

This extremely simple, but powerful technique is called the Premortem method of risk assessment.  The technique was coined and promoted by Gary Klein, a research psychologist and an author.  It is generally thought of as a Project Management technique but it can be applied to Explaining Analytics just as well – specifically to the presentation phase.

Consider the conditions of a typical project starting a new endeavor, including a typical analytic project.  The planning and the discussions are all on the potential of the positive outcome and how it can be used.  Assume for good measure that the project has excellent upper management support.  Under such a scenario no one wants to predict a failure.  Human nature and organizational pressure keeps one from speaking up and from being “the one” that is the negative.  But, of course, projects fail at a spectacular rate.  In fact, analytical projects may fail at an even greater rate.  Many analytical projects represent new techniques for the organization, new data sources and new ways of doing business - all the attributes that make this new field of analytics exciting but yet risky.  My experience has shown that analytic projects may face more inertia than other projects and frequently have a high hurdle to get started.  There are more unknowns, or perhaps more perceived unknowns, by others with less experience with analytics.  Discussions of all that can go wrong adds to that inertia. 

Pre-launch risk assessment efforts are hindered by the reluctance to discuss failure.  In such an environment, you may not want to risk looking like the naysayer, the negative one, the overly cautious.

In the Premortem approach as presented by Dr. Klein, the process works differently than a traditional risk assessment.  Before the project begins, the project members have been told to consider that the project has failed.  In fact, it has failed spectacularly.  Under this assumed scenario the project members are to take two minutes to write down the reasons they can think of for why it failed.  The project leader then goes around the room and ask each team member to read a different reason why it failed from their lists.  The list is then reviewed and corrective active taken, if necessary.

The power of this technique is the negative statements are not given as risks, or concerns to be punched down by overinvested supporters of the project.  In this scenario, the negative statements are seen as creative predictions of a future outcome.

On hearing this process explained, not only is the value in project planning and management apparent but the value in preparation for a presentation is obvious.  Although explaining analytics requires many steps, there is no doubt that a presentation is often the most crucial stage in presenting analytical findings; sometimes it is your only chance in front of an audience.

Typically, such planning includes:

  • what findings and conclusions you want drawn from the presentation;
  • what decisions and what actions you want from the work;
  • how you see the work and what you think you have found.

Anyone who has been through this many times knows it may not turn out the way you expect.

The likelihood of a success can be enhanced by a version of a Premortem.  With the draft of the presentation before the project team, the leader can claim the presentation failed.  The work was not understood, no positive decisions were made, and frankly, many left confused and wondering if there really is any value in analytics for your company.

Any experienced analyst can think of several potential items that might be listed including:  unexpected attendees, debated data quality, different assumptions, different objectives, confusion in interpreting specific slides, short attention spans.  With a draft of the presentation in front of you, the list could include quite detailed issues.

It is easy to see how a 5 minute preparation talk painting the future failure presentation scenario, 2 minutes of writing the reasons for the failure and 15 minutes of discussion can easily save a presentation and maybe the project dependent on the presentation being accepted.  A small investment of time for a potentially big payoff.

The saying goes, “In school you are given the lessons, then take the test. In life, you take the test, then are given the lessons.”  The Premortem technique turns the tables.

You can learn more of these skills in my Business Knowledge Series course, Explaining Analytics to Decision Makers: Insights to Action.

Explaining analytics: Learning before failure #SAStraining Click To Tweet

About Author

Jeff Zeanah

President of Z Solutions, Inc.

Jeff Zeanah is President of Z Solutions, Inc., a firm focused on customized analytical consulting for over 20 years. He has consulted with industry leaders in manufacturing, retail, software, public health, science, finance, and utilities. A frequent lecturer on the topic of explaining analytics and the management of analytical projects, Jeff enjoys sharing field experiences with colleagues. As a recognized expert on neural networks and a broad range of exploratory data mining tools, Jeff has authored papers on neural networks, exploratory data mining, and the implementation of those techniques in organizations. He is the developer of exploratory approaches and techniques that have been used by Fortune 500 companies, independent researchers, government agencies, and over 30 universities worldwide. Jeff’s approaches have been applied in areas as diverse as market research, software license compliance, tasting wines, nutrition, sizing electric transformers, and classifying stars.

Related Posts

Comments are closed.

Back to Top