Eric Siegel: An interview by Bob Morris

SiegedlEric Siegel, PhD, founder of Predictive Analytics World and Text Analytics World, and Executive Editor of the Predictive Analytics Times.com, makes the how and why of predictive analytics understandable and captivating. In addition to being the author of Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, he is a former Columbia University professor who used to sing educational songs to his students, and a renowned speaker, educator and leader in the field. He has appeared on Bloomberg TV and Radio, Fox News, BNN (Canada), Israel National Radio, Radio National (Australia), The Street, Newsmax TV, and NPR affiliates. Eric and his book have been featured in BusinessWeek, CBS MoneyWatch, The Financial Times, Forbes, Forrester, Fortune, The Huffington Post, The New York Times, The Seattle Post-Intelligencer, The Wall Street Journal, The Washington Post, Wall Street Journal, and MarketWatch.

* * *

Morris: Before discussing Predictive Analytics, a few general questions. Years ago, was there a turning point (if not an epiphany) that set you on the career course you continue to follow? Please explain.

Siegel: Yes, it was in 1991, my summer between college and Columbia’s doctoral program, when I realized I wanted to pursue machine learning: the ability for computers to learn from experience/examples (aka, data!), which is at the heart of predictive analytics. With this technology, the computer pours through examples to learn how to PREDICT. (A big contribution that summer was touring Vancouver with my old buddy Alex Chaffee, who whispered into my ear some magic words he’d learned at Reed College on the topic.)

Morris: What do you know now about the business world that you wish you knew when you when to work full-time for the first time? Why?

Siegel: Innovative technology doesn’t sell itself, and its sale is dependent more on a gradual social process (or “social experiment?”) than on a perfectly-written speech or white paper, no matter how well put together the pitch is.

Morris: First, who has had the greatest impact on your professional development? How so?

Siegel: I’m so rarely asked that question! I like to think my main strength is the ability to explain a technology to any audience (specifically, with the most passion, topics within my very favorite field: predictive analytics). Looking back, I can truly see how so many enthusiastic teachers in technical fields as well as various humanities – from high school on up – contagiously infected me with what it means to effectively 1) understand, 2) find excitement in, and 3) communicate. It’s kind of a particular socialization process.

So, in answering your question, instead of deifying one person, I’ll deify the education system! This includes public schools in Vermont, Brandeis University, and Columbia University (where I was later on the faculty), where I was lucky enough to have too many strong mentors to count (see my book’s acknowledgements for a few of them).

Morris: Here are several of my favorite quotations to which I ask you to respond. First, from Lao-Tzu’s Tao Te Ching:

“Learn from the people

Plan with the people

Begin with what they have

Build on what they know

Of the best leaders

When the task is accomplished

The people will remark

We have done it ourselves.”

Siegel: Yes – a consultant’s job is often to make his assistance seem only supportive of the client’s successful execution.

Morris: Next, from Voltaire: “Cherish those who seek the truth but beware of those who find it.”

Siegel: Ah, yes, humility is important. In fact, I’ve been working on my humility and I think I’ve got it down better than anyone I know. (That is meant to be humorous.)

Morris: Finally, from Peter Drucker: “There is surely nothing quite so useless as doing with great efficiency what should not be done at all.”

Siegel: Checks and balances before triggering a project!

Morris: In Tom Davenport’s latest book, Judgment Calls, he and co-author Brooke Manville offer “an antidote for the Great Man theory of decision making and organizational performance”: organizational judgment. That is, “the collective capacity to make good calls and wise moves when the need for them exceeds the scope of any single leader’s direct control.” What do you think?

Siegel: Beyond again invoking “checks and balances,” I’ll also tie this to predictive analytics specifically: There are too many tactical decisions to manage manually, so it is the collective capacity of DATA that will inform each one, thus tweaking the aggregate effectiveness of so many tactical decisions.

Morris: The greatest leaders throughout history (with rare exception) were great storytellers. What do you make of that?

Siegel: Nice point. Change comes of monitoring and managing a gradual social shift. Each such cycle could be a novel.

Morris: Most change initiatives either fail or fall far short of original (perhaps unrealistic) expectations. More often than not, resistance is cultural in nature, the result of what James O’Toole so aptly characterizes as “the ideology of comfort and the tyranny of custom.”

Siegel: Yes, you see the social process, as people are ambivalently shepherded through change — change that they fear so greatly and have build walls of rationalization against.

Morris: Now please shift your attention to Predictive Analytics. Why did you decide to write it?

Siegel: Beyond fully setting forth the concepts for business people of all backgrounds who wish to gain value from predictive analytics, I wanted engage, entertain, and enlighten even non-business lay-readers in a “pop science” kind of style, to share with the world at large what is fascinating about this advanced technology and the potential it holds to make the world a better place in so many ways.

Morris: Were there any head-snapping revelations while writing it? Please explain.

Siegel: Before writing the book, I caught wind of IBM’s Watson computer, which succeeded on the TV quiz show “Jeopardy!” by applying the same core technology to predict, for a given question/answer pair, “Will this turn out to be the right answer?” Wow! I devoted a chapter to that. Also, late 2012, after completing the final chapter on an advanced method of persuasion analytics that renders marketing more influential – just about to finalize the complete manuscript – I caught wind of the fact that the Obama campaign had in fact used that very particular form of predictive analytics (in addition to their other various analytics work that got plenty of press). I interviewed them for the full story and was able to insert a lengthy sidebar on it into the book at the last second.

Morris: What are the most common misconceptions about predictive analytics (PA)?

Siegel: There’s a tie between two big misconceptions:

1. Predictive analytics is the same as forecasting. But no — instead of foreseeing general trends like overall sales or the direction of the economy, predictive analytics renders a prediction for each individual. In that way it improves mass-scale operations, driving them one micro-decision at a time.

2. Predictive analytics is difficult to understand.

Morris: In essence, what is “the art of learning”?

Siegel: Whether you’re conceiving of your own learning process, or developing a learning process for a computer to follow, learning depends on oversimplification. You cannot jump from examples (even millions of examples) to a conclusion without making an assumption about the world that in some way simplifies. Learning is never perfect. The art is in defining that simplification. That sounds pretty abstract but if you look through Chapter 4 — the chapter that focuses on learning  —  you’ll see the detailed actualities are in fact easy to follow.

Morris: Why must a computer learn in order to predict?

Siegel: Looking at a situation and putting odds on the future outcome depends on already having certain insight (or intelligence, if you like that word), insight which must be earned by a learning process. What’s learned is the pattern, rule, or equation that takes a new predictive question that’s never been seen before – such as, “Will this particular customer in this situation buy the product?” – and assigns a predictive score (probability). That pattern/rule/equation is called a predictive model and is best refined automatically over many examples (data). That’s a learning process.

Morris: Why expedite installation of a predictive model within a field operation? What are the potential benefits of doing that?

Siegel: Acting on predictions combats financial risk, fortifies healthcare, conquers spam, toughens crime fighting, and boosts sales.

Morris: What actually makes data predictive?

Siegel: Data always encodes all sorts of predictive relationships (correlations), such as “Earlier retirees have lower life expectancies.”

Morris: How does prediction transform risk to opportunity?

Siegel: With risk, you’re talking about negative outcomes. If you put the odds on a large number of possible negative outcomes, such as the chance each insurance policyholder will have a car accident, the risk can be pooled and managed more effectively.

Morris: What kind of predictive model can be understood by almost everyone?

Siegel: Decision trees! This form of predictive model is just a bunch of if-then statements, like business rules. But it is a more adept than you might expect.

Morris: What key innovation in predictive analytics has crowdsourcing helped to develop?

Siegel: Advancements in the increasingly popular “ensemble model,” which is a way to group a bunch of simpler predictive models together into a single, cooperating team, emerged from crowdsourcing – specifically, public predictive analytics competitions.

Morris: How is a person like a quantum particle?

Siegel: Marketing and healthcare treatments are meant to INFLUENCE, and analytics is meant to optimize said influence. But whether a person can be influenced (to buy, or to improve healthwise) by a certain treatment is in fact impossible to know with entire certainty, just as it is impossible to fully know both the spin and velocity of a quantum particle.

Morris: Here is a question I have been eager to ask you since the first reading of your brilliant book. (I hope I express it properly!) To what extent is the future changed by predictions made about it before it occurs?

Siegel: A lot! That is the whole point of predictive analytics – to empower an organization to improve operations in order to make more positive change tomorrow. So, when we predict a customer will buy if contacted, then the organization chooses to contact them (no brainer!). We’ve just landed a new sale for tomorrow. If we predict a patient will die, the provider is signaled to consider intervening. We’ve just saved a life tomorrow.

Morris: To what extent (if any) have reactions to the book since publication been different from your expectations, if not predictions? Please explain.

Siegel: Most of the reactions have been in alignment with the kind of hopes and dreams an author has during all those late nights writing. I get a huge kick out of finding ways to show an “arcane” technology is not weird, difficult, or boring — and is in fact super-relevant to everyone. And so it’s very gratifying that some believe I’ve successfully done so.

Morris: Of all the five effects of prediction, which seems to have the greatest impact? Why?

Siegel: The five Effects are: the Prediction, Induction, Data, Persuasion, and Ensemble Effects. I would say the first is most fundamental and powerful. The Predictive Effect states that a little prediction goes a long way – predicting better than guessing, even if not accurately, serves to tip the balance in the numbers games all organizations play and deliver a great boost in aggregate effectiveness and your bottom line numbers.

Morris: In your opinion, what is the single worst mistake that companies make when attempting to implement a PA initiative?

Siegel: Beginning the hands-on analytical execution before gaining buy-in for its eventual deployment (i.e., use) in operations.

Morris: Let say that a CEO has read and then (hopefully) re-read Predictive Analytics and is now determined to improve decision-making capabilities based on predictive analytics — at all levels and in all areas of the given enterprise. Where to begin?

Siegel: Beyond the requisite background given by a book like mine, we’ve put together a list of resources to get started with the how-to and execution: The Predictive Analytics Guide at www.pawcon.com/guide.

Morris: For more than 25 years, it has been my great pleasure as well as privilege to work closely with the owner/CEOs of hundreds of small companies, those with $20-million or less in annual sales. In your opinion, of all the material you provide in Predictive Analytics, which do you think will be of greatest value to leaders in small companies? Please explain.

Siegel: It all applies. The benefit of predictive analytics depends not on organizational size per se, but on there being a large-scale operation to optimize, such as direct mail to a list of 30,000 or more, and a proportionally large data set from which to learn.

 

Morris: Which question had you hoped to be asked during this interview — but weren’t — and what is your response to it?

Siegel: What do I spend most of my work time on? Answer: Growing the conference series, Predictive Analytics World, which is the leading cross-vendor event focused on commercial deployment (www.pawcon.com).

* * *

Eric cordially invites you to check out the resources at these websites:

Predictive Analytics link

Predictive Analytics World link

Eric’s Amazon page link

Posted in

Leave a Comment





This site uses Akismet to reduce spam. Learn how your comment data is processed.