KDnuggets Home » News » 2014 » May » Opinions, Interviews » Interview: Richard Wendell, VP, Data Science, TE Connectivity on Strategy for Analytics Projects ( 14:n13 )

Interview: Richard Wendell, VP, Data Science, TE Connectivity on Strategy for Analytics Projects


We discuss the last mile of the execution path of Analytics projects, five critical pillars of success and data-driven decision making through advanced analytics.



Richard WendellRichard Wendell is the Vice President, Data Science and Strategic Analytics for TE Connectivity (TEL), the $13.5B global electronics manufacturer. In this role, Mr. Wendell leads the global team responsible for data science and analytics across the company. Mr. Wendell was brought into the company to construct the data science team from scratch and to pioneer the company’s move into advanced analytics.

Prior to joining TE Connectivity, Mr. Wendell was Vice President, Global Strategy & Business Development at American Express, where he led the company’s move into analytics-driven business models. Before American Express, Mr. Wendell's experience include co-founding several technology start-up companies, driving the successful turnaround of an IT services firm, and management consulting to Fortune 100 telecommunications clients. Mr. Wendell holds an MBA in Decision Technologies from the Tepper School at Carnegie Mellon and an undergraduate degree in Philosophy and Mathematics, Summa Cum Laude, from Brandeis University. Mr. Wendell is a regular speaker at events on big data and innovation.

Here is my interview with him:

Anmol Rajpurohit Q1. Why do you believe that most of the Analytics projects fail during the execution of "last mile"? Does this mean that the bigger problems related to Analytics projects are related to business and not technology?

Richard Wendell: Yes, I get this question a lot.  As I mentioned in my presentation at the Big Data Innovations Summit, MIT quoted 55% of Big Data projects as failing to drive the desired outcome.  By most testing standards, 45% passing rate is a failing score.  This result is a terrible thing for those of us who are in the field.  We need to get this score up!  The question is: where do we start?

From my experience, primary causes of such project failure happen after the team performs good analysis and delivers powerful analytics insight.  For some reason, the insight, however brilliant, often doesn’t get acted on by the business.  How many times have you heard an analyst say, “Well, I proved to them that they need to do such and such differently…but they didn’t listen!”  At the end of the day, the objective in a company is to make money.  I see this phase, between the time when an insight is delivered and the time that money is made, as the last mile of Analytics engagements.

Last mile aheadFrom my experience, most projects fail in the last mile.  They do not fail because someone used the wrong algorithm, modeling technique or didn’t have enough data.  Yet, most of those in our profession are focused on creating better algorithms, modeling techniques or getting better data.  Of course we need focus on those things too.  But what if the root causes for project failure are rooted in other issues that we are not focusing on?  Then we will never get our profession’s failing score up.

I believe that most analytics and data science projects fail in the last mile because of what I see as a lack in the team’s “Analytics EQ,” or Emotional Intelligence.  Instilling more focus into a team’s Analytics EQ is what my five pillars are about.


AR: Q2. The five critical pillars of success that you highlighted seem to be very apt for top-down strategy, i.e. executives and managers using those pillars to achieve success. From the bottom-up perspective, what do you consider as the important pillars of success (would the same 5 apply?) ? In other words, how can the data scientists and data analysts contribute towards ensuring the success of their project?

RW: For starters, for readers who haven’t heard my presentation yet, the

Five Pillars of Success for Data Science teams are:
  1. Align analytics engagements with the corporate strategy
  2. Ignite stakeholder engagement
  3. Sharpen the team’s focus
  4. Drive change management
  5. Recruit and mentor key talent

 
Five pillars of successIf a data science team pays attention to these five areas, my experience is that the success rate should be better than 45%.  Now, I won’t argue that these are the only pillars that contribute to success.  I would be delighted if people in the community could suggest other pillars.  My hope in putting these five pillars out is that we can give the data science community a few manageable areas of focus for getting their team’s EQ up, which will drive better engagement success.

I don’t see these areas of focus as top down.  I see the pillars as critical items that aren’t mentioned as part of CRISP-DM. 
If every single analyst on a project can’t articulate how the project aligns with the corporate strategy, then that is a problem.  If every single analyst can’t tell me how their project helps a key stakeholder get a larger bonus, then that is a problem.  The entire team needs to be engaged in these issues, not just the top leadership. 
My analysts are critical to helping recruit and mentor new talent.  Of course, there is a leadership component as well.  It is the job of analytics leaders, whether it be in a start-up, a consulting firm or a fortune 500, to ensure that everyone on the team is connecting the dots.  Not having the right context to connect the dots is where things start to go off the rails.  This usually happens when well-meaning analysts, lacking adequate context, make what they see as reasonable assumptions.  These assumptions can spell death for an analytics or data science engagement.  Leaders need to set a culture where analysts understand that it is incumbent on them to escalate quickly to obtain critical context.

AR: Q3. Has the advanced analytics truly pushed the envelope for data-driven decision making or we still have most of decisions at top level being made based on just intuition? Can you share a use case from your experience where you were able to propose and achieve data-driven decision making through advanced analytics?

RW: I agree with Brynjolfsson that most decisions at the top are largely made by HIPPO’s (Highest Paid Person’s Opinion).  That said, many decisions at the top have evolved into something so complex that it becomes next to impossible to find the data that informs a “right” answer.  A few years back, people were saying that Harrah’s would fire a senior leader who ran an experiment without a control group.  Even if this is true, Harrah’s is in the minority.

Many complex decisions at the top occur because Analysts, Managers and Directors have failed to apply a data-driven approach when the decision was on a smaller scale.  So the decision either isn’t made or is made sub optimally. Data-driven decision-making If this occurs in a critical area in the business, eventually the situation snowballs and results in a much more complex decision—which may be tough to dissect with data.  In that regard, I don’t believe that applying data-driven decision making at the top is the right starting point.  My focus is about empowering data-driven decision making more broadly within the organization.

To some extent, I think the proper application of business experimentation, data and analytics can enable decisions to be made less centrally, less top-down (decentralized).

Second and last part of interview: Richard on Role of Analytics in Organizations

Related: