Posts tagged: Flaw of Averages

11 Ways Humans Kill Good Analysis

Failure to CommunicateIn my last post, I talked about the immense value of FAME in analysis (Focused, Actionable, Manageable and Enlightening). Some of the comments on the post and many of the email conversations I had regarding the post sparked some great discussions about the difficulties in achieving FAME. Initially, the focus of those discussions centered on the roles executives, managers and other decisions makers play in the final quality of the analysis, and I was originally planning to dedicate this post to ideas decision makers can use to improve the quality of the analyses they get.

But the more I thought about it, the more I realized that many of the reasons we aren’t happy with the results of the analyses come down to fundamental disconnects in human relations between all parties involved.

Groups of people with disparate backgrounds, training and experiences gather in a room to “review the numbers.” We each bring our own sets of assumptions, biases and expectations, and we generally fail to establish common sets of understanding before digging in. It’s the type of Communication Illusion I’ve written about previously. And that failure to communicate tends to kill a lot of good analyses.

Establishing common understanding around a few key areas of focus can go a long way towards facilitating better communication around analyses and consequently developing better plans of action to address the findings.

Here’s a list of 11 key ways to stop killing good analyses:

  1. Begin in the beginning. Hire analysts not reporters.
    This isn’t a slam on reporters, it’s just recognition that the mindset and skill set needed for gathering and reporting on data is different from the mindset and skill set required for analyzing that data and turning it into valuable business insight. To be sure, there are people who can do both. But it’s a mistake to assume these skill sets can always be found in the same person. Reporters need strong left-brain orientation and analysts need more of a balance between the “just the facts” left brain and the more creative right brain. Reporters ensure the data is complete and of high quality; analysts creatively examine loads of data to extract valuable insight. Finding someone with the right skill sets might cost more in payroll dollars, but my experience says they’re worth every penny in the value they bring to the organization.
  2. Don’t turn analysts into reporters.
    This one happens all too often. We hire brilliant analysts and then ask them to spend all of their time pulling and formatting reports so that we can do our own analysis. Everyone’s time is misused at best and wasted at worst. I think this type of thing is a result of the miscommunication as much as a cause of it. When we get an analysis we’re unhappy with, we “solve” the problem by just doing it ourselves rather than use those moments as opportunities to get on the same page with each other. Web Analytics Demystified‘s Eric Peterson is always saying analytics is an art as much as it is a science, and that can mean there are multiple ways to get to findings. Talking about what’s effective and what’s not is critical to our ultimate success. Getting to great analysis is definitely an iterative process.
  3. Don’t expect perfection; get comfortable with some ambiguity
    When we decide to be “data-driven,” we seem to assume that the data is going to provide perfect answers to our most difficult problems. But perfect data is about as common as perfect people. And the chances of getting perfect data decrease as the volume of data increases. We remember from our statistics classes that larger sample sizes mean more accurate statistics, but “more accurate” and “perfect” are not the same (and more about statistics later in this list). My friend Tim Wilson recently posted an excellent article on why data doesn’t match and why we shouldn’t be concerned. I highly recommend a quick read. The reality is we don’t need perfect data to produce highly valuable insight, but an expectation of perfection will quickly derail excellent analysis. To be clear, though, this doesn’t mean we shouldn’t try as hard as we can to use great tools, excellent methodologies and proper data cleansing to ensure we are working from high quality data sets. We just shouldn’t blow off an entire analysis because there is some ambiguity in the results. Unrealistic expectations are killers.
  4. Be extremely clear about assumptions and objectives. Don’t leave things unspoken.
    Mismatched assumptions are at the heart of most miscommunications regarding just about anything, but they can be a killer in many analyses. Per item #3, we need to start with the assumption that the data won’t be perfect. But then we need to be really clear with all involved what we’re assuming we’re going to learn and what we’re trying to do with those learnings. It’s extremely important that the analysts are well aware of the business goals and objectives, and they need to be very clearly about why they’re being asked for the analysis and what’s going to be done with it. It’s also extremely important that the decision makers are aware of the capabilities of the tools and the quality of the data so they know if their expectations are realistic.
  5. Resist numbers for number’s sake
    Man, we love our numbers in retail. If it’s trackable, we want to know about it. And on the web, just about everything is trackable. But I’ll argue that too much data is actually worse than no data at all. We can’t manage what we don’t measure, but we also can’t manage everything that is measurable. We need to determine which metrics are truly making a difference in our businesses (which is no small task) and then focus ourselves and our teams relentlessly on understanding and driving those metrics. Our analyses should always focus around those key measures of our businesses and not simply report hundreds (or thousands) of different numbers in the hopes that somehow they’ll all tie together into some sort of magic bullet.
  6. Resist simplicity for simplicity’s sake
    Why do we seem to be on an endless quest to measure our businesses in the simplest possible manner? Don’t get me wrong. I understand the appeal of simplicity, especially when you have to communicate up the corporate ladder. While the allure of a simple metric is strong, I fear overly simplified metrics are not useful. Our businesses are complex. Our websites are complex. Our customers are complex. The combination of the three is incredibly complex. If we create a metric that’s easy to calculate but not reliable, we run the risk of endless amounts of analysis trying to manage to a metric that doesn’t actually have a cause-and-effect relationship with our financial success. Great metrics might require more complicated analyses, but accurate, actionable information is worth a bit of complexity. And quality metrics based on complex analyses can still be expressed simply.
  7. Get comfortable with probabilities and ranges
    When we’re dealing with future uncertainties like forecasts or ROI calculations, we are kidding ourselves when we settle on specific numbers. Yet we do it all the time. One of my favorite books last year was called “Why Can’t You Just Give Me the Number?” The author, Patrick Leach, wrote the book specifically for executives who consistently ask that question. I highly recommend a read. Analysts and decision makers alike need to understand the of pros and cons of averages and using them in particular situations, particularly when stacking them on top of each other. Just the first chapter of the book Flaw of Averages does an excellent job explaining the general problems.
  8. Be multilingual
    Decision makers should brush up on basic statistics. I don’t think it’s necessary to re-learn all the formulas, but it’s definitely important to remember all the nuances of statistics. As time has passed from our initial statistics classes, we tend to forget about properly selected samples, standard deviations and such, and we just remember that you can believe the numbers. But we can’t just believe any old number. All those intricacies matter. Numbers don’t lie, but people lie, misuse and misread numbers on a regular basis. A basic understanding of statistics can not only help mitigate those concerns, but on a more positive note it can also help decision makers and analysts get to the truth more quickly.

    Analysts should learn the language of the business and work hard to better understand the nuances of the businesses of the decision makers. It’s important to understand the daily pressures decision makers face to ensure the analysis is truly of value. It’s also important to understand the language of each decision maker to shortcut understanding of the analysis by presenting it in terms immediately identifiable to the audience. This sounds obvious, I suppose, but I’ve heard way too many analyses that are presented in “analyst-speak” and go right over the heard of the audience.

  9. Faster is not necessarily better
    We have tons of data in real time, so the temptation is to start getting a read almost immediately on any new strategic implementation, promotion, etc. Resist the temptation! I wrote a post a while back comparing this type of real time analysis to some of the silliness that occurs on 24-hour news networks. Getting results back quickly is good, but not at the expense of accuracy. We have to strike the right balance to ensure we don’t spin our wheels in the wrong direction by reacting to very incomplete data.
  10. Don’t ignore the gut
    Some people will probably vehemently disagree with me on this one, but when an experienced person says something in his or her gut says something is wrong with the data, we shouldn’t ignore it. As we stated in #3, the data we’re working from is not perfect so “gut checks” are not completely out of order. Our unconscious or hidden brains are more powerful and more correct than we often give them credit for. Many of our past learnings remain lurking in our brains and tend to surface as emotions and gut reactions. They’re not always right, for sure, but that doesn’t mean they should be ignored. If someone’s gut says something is wrong, we should at the very least take another honest look at the results. We might be very happy we did.
  11. Presentation matters a lot.
    Last but certainly not least, how the analysis is presented can make or break its success. Everything from how slides are laid out to how we walk through the findings matter. It’s critically important to remember that analysts are WAY closer to the data than everyone else. The audience needs to be carefully walked through the analysis, and analysts should show their work (like math proofs in school). It’s all about persuading the audience and proving a case and every point prior to this one comes into play.

The wealth and complexity of data we have to run our businesses is often a luxury and sometimes a curse. In the end, the data doesn’t make our businesses decisions. People do. And we have to acknowledge and overcome some of our basic human interaction issues in order to fully leverage the value of our masses of data to make the right data-driven decisions for our businesses.

What do you think? Where do you differ? What else can we do?

Why most sales forecasts suck…and how Monte Carlo simulations can make them better

Sales forecasts don’t suck because they’re wrong.  They suck because they try to be too right. They create an impossible illusion of precision that ultimately does a disservice to managers who need accurate forecasts to assist with our planning. Even meteorologists — who are scientists with tons of historical data, incredibly high powered computers and highly sophisticated statistical models — can’t forecast with the precision we retailers attempt to forecast. And we don’t have nearly the data, the tools or the models meteorologists have.

Luckily, there’s a better way. Monte Carlo simulations run in Excel can transform our limited data sets into statistically valid probability models that give us a much more accurate view into the future. And I’ve created a model you can download and use for yourself.

There are literally millions of variables involved in our weekly sales, and we clearly can’t manage them all. We focus on the few significant variables we can affect as if they are 100% responsible for sales, but they’re not and they are also not 100% reliable.

Monte Carlo simulations can help us emulate real world combinations of variables, and they can give us reliable probabilities of the results of combinations.

But first, I think it’s helpful to provide some background on our current processes…

We love our numbers, but we often forget some of the intricacies about numbers and statistics that we learned along the way. Most of us grew up not believing a poll of 3,000 people could predict a presidential election. After all, the pollsters didn’t call us. How could the opinions of 3,000 people predict the opinions of 300 million people?

But then we took our first statistics classes. We learned all the intricacies of statistics. We learned about the importance of properly generated and significantly sized random samples. We learned about standard deviations and margins of errors and confidence intervals. And we believed.

As time passed, we moved on from our statistics classes and got into business. Eventually, we started to forget a lot about properly selected samples, standard deviations and such and we just remembered that you can believe the numbers.

But we can’t just believe any old number.

All those intricacies matter. Sample size matters a lot, for example. Basing forecasts, as we often do, on limited sets of data can lead to inaccurate forecasts.

Here’s a simplified explanation of how most retailers that I know develop sales forecasts:

  1. Start with base sales from last year for the the same time period you’re forecasting (separating out promotion driven sales)
  2. Apply the current sales trend (which is maybe determined by an average of the previous 10 week comps). This method may vary from retailer to retailer, but this is the general principle.
  3. Look at previous iterations of the promotions being planned for this time period. Determine the incremental revenue produced by those promotions (potentially through comparisons to control groups). Average of the incremental results of previous iterations of the promotion, and add that average to the amount determined in steps 1 and 2.
  4. Voilà! This is the sales forecast.

Of course, this number is impossibly precise and the analysts who generate it usually know that. However, those on the receiving end tend to assume it is absolutely accurate and the probability of hitting the forecast is close to 100% — a phenomenon I discussed previously when comparing sales forecasts to baby due dates.

As most of us know from experience, actually hitting the specific forecast almost never happens.

We need accuracy in our forecasts so that we can make good decisions, but unjustified precision is not accuracy. It would be far more accurate to forecast a range of sales with accompanying probabilities. And that’s where the Monte Carlo simulation comes in.

Monte Carlo simulations

Several excellent books I read in the past year (The Drunkard’s Walk, Fooled by Randomness, Flaw of Averages, and Why Can’t You Just Give Me a Number?) all promoted the wonders of Monte Carlo simulations (and Sam Savage of Flaw of Averages even has a cool Excel add-in). As I read about them, I couldn’t help but think they could solve some of the problems we retailers face with sales forecasts (and ROI calculations, too, but that’s a future post). So I finally decided to try to build one myself. I found an excellent free tutorial online and got started. The results are a file you can download and try for yourself.

A Monte Carlo simulation might be most easily explained as a “what if” model and sensitivity analysis on steroids. Basically, the model allows us to feed in a limited set of variables about which we have some general probability estimates and then, based on those inputs, generate a statistically valid set of data we can use to run probability calculations for a variety of possible scenarios.

It turns out to be a lot easier than it sounds, and this is all illustrated in the example file.

The results are really what matters. Rather than producing a single number, we get probabilities for different potential sales that we can use to more accurately plan our promotions and our operations. For example, we might see that our base business has about a 75% chance of being negative, so we might want to amp up our promotions for the week in order have a better chance of meeting our growth targets.  Similarly, rather than reflexively “anniversaring” promotions, we can easily model the incremental probabilities of different promotions to maximize both sales and profits over time.

The model allows for easily comparing and contrasting the probabilities of multiple possible options. We can use what are called probability weighted “expected values” to find our best options. Basically, rather than straight averages that can be misleading, expected values are averages that are weighted based on the probability of each potential result.

Of course, probabilities and ranges aren’t as comfortable to us as specific numbers, and using them really requires a shift in mindset. But accepting that the future is uncertain and planning based on the probabilities of potential results puts us in the best possible position to maximize those results. Understanding the range of possible results allows for better and smarter planning. Sometimes, the results will go against the probabilities, but consistently making decisions based on probabilities will ultimately earn the best results over time.

One of management’s biggest roles is to guide our businesses through uncertain futures. As managers and executives, we make the decisions that determine the directions of our companies. Let’s ensure we’re making our decisions based on the best and most accurate information — even if it’s not the simplest information.

What do you think? What issues have you seen with sales forecasts? Have you tried my example? How did it work for you?

Best Business Books of the Year

With the holiday season upon us, I thought I would write about my favorite business books of the year to provide some gift giving ideas for you and your teams. Here, in no particular order, are my favorites among the books I read this year. (Note: These books were not all published this year, but since I read them this year I’m including them in my list.)

Six Pixels of Separation: Everyone is Connected. Connect Your Business to Everyone.
by Mitch Joel

Six Pixels of Separation begins as a primer for any business leader with limited knowledge of the Internet’s capabilities and quickly turns into an indispensable set of guidelines and advice for any business person who plans to make use of the web (which should be any business person). Mitch Joel offers excellent insight and plenty of simple, direct, digestible advice. This is a must read.

The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty
by Sam L. Savage

Every business person should read this book. We are so often looking for precise numbers when precise numbers are unrealistic. The reality is, we would actually be much more accurate to use probabilities and ranges when referencing uncertain number such as sales forecasts or project timelines. Savage takes us through the dangers of using averages to describe distributions and offers solid solutions that can be used to better manage our business.
Preview Flaw of Averages

Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets
by Nassim Nicholas Taleb

This book made me think more than any book in recent memory. That may be partly because it’s pretty dense and I had to read it more slowly than I normally read. However, I’ll give a lot more credit to the fact that Taleb’s makes some very interesting points about the amount of randomness in our lives and how that randomness is all too often mistaken for something more substantive.
Preview Fooled by Randomness

How We Decide
by Jonah Lehrer
I loved this book. Jonah Lehrer takes us through some fairly common behavior economics principles and experiments, but the very interesting twist he takes is to explain the brain mechanics that drive our thinking and decisions. He really uncovers why we’re “predictably irrational” and provides great insight into how we make decisions and how we can use that knowledge to improve our decision making.

The Drunkard’s Walk: How Randomness Rules Our Lives
by Leonard Mlodinow

I’m on a randomness kick lately, and this is the book that got me started on it. Mlodinow does a nice job of illustrating some of the finer statistical points in a pretty accessible manner. While this book isn’t as deep at the book I’m currently reading, “Fooled by Randomness,” it’s definitely an easier read and does a nice job of covering the basics.
Preview The Drunkard’s Walk

Sway: The Irresistible Pull of Irrational Behavior
by Ori Brafman, Rom Brafman

Another one of the behavior economics books I so love. This one has some pretty interesting stories and anecdotes, and its insights benefit from one of the writers being a psychologist and the other a businessman.
Preview Sway

More Than a Motorcycle: The Leadership Journey at Harley-Davidson
By Rich Teerlink and Lee Ozley

This is a very interesting book about culture change at Harley-Davidson during the ’90s written by the CEO and lead consultant who initiated the change. It can be a bit dry at times, but the details behind the thinking and the execution are excellent. I learned a lot by reading it.
Preview More than a Motorcycle


And here are some great books that I re-read this year:

The OPEN Brand: When Push Comes to Pull in a Web-Made World
by Kelly Mooney, Nita Rollins
The world is changing rapidly, and those who fail to realize it will be left in the dust. However, those who open their brand and see the value of allowing their best customers to participate in the brand will not only reap the benefits of those customers ideas, but they will also benefit from those customers becoming the largest and more credible Marketing department a company could have. Kelly Mooney and Nita Rollins explore these themes in an extremely insightful book that comes with lots of examples that help the reader visualize how these ideas could apply to his or her own business. The writing style and formatting is fun and extremely easy to read. This is a great handbook for any marketer in the 21st century.

Moneyball: The Art of Winning an Unfair Game
by Michael Lewis

While this is ostensibly a baseball book about the success of Oakland A’s GM Billy Beane, I actually found this to be an excellent business book. Michael Lewis tells the story of Beane defying the conventional wisdom of longtime baseball scouts about what good baseball players look like. Rather than trust scouts who literally would determine a baseball player’s prospects by how he physically looked, Beane went to the data as a disciple of Bill James’ Sabermetrics theories. Lewis describes how James took a new look at traditional baseball statistics and created new statistics that were actually more causally related to winning games. By following the James’ approach, Beane was able to put together consistently winning teams while working with one of the lowest payrolls in the Major Leagues. How can the same principles of trusting data over tradition and “gut” play in the business world? That is a thought I constantly ponder thanks to reading this book.
Preview Moneyball

The Culture Code: An Ingenious Way to Understand Why People Around the World Live and Buy as They Do
by Clotaire Rapaille

I picked this book up on a whim one day because the title was interesting. I was quickly engrossed by reading the story in the introduction of Clotaire Rapaille’s work with Chrysler on Jeep Wrangler. He describes the “code” word for Jeep in America is HORSE and advises executives to design round headlights instead of square headlights because horses have round eyes. They think he’s nuts, of course, but when it turns out round headlights are cheaper they go with them — and they’re a hit. They also then position the Wrangler as a “horse” in their ads and have great success. Rapaille goes on to describe what he means by “culture code” and details some of the hidden cultural patterns that affect most all of us. Some samples of other codes within the book are:
– The American Culture Code for love is FALSE EXPECTATION
– The female code for sex is VIOLENCE (Whoa! You’ve got to read the book to understand)
– The code for hospital in America is PROCESSING PLANT

There are tons more of these interesting observations embedded in short, easy-to-read chapters. Whether or not you buy into everything he says, it’s very interesting to see how he developed each code and certainly will expand your understanding of how and why people behave as they do under the powerful forces of culture
Preview The Culture Code

Predictably Irrational: The Hidden Forces That Shape Our Decisions
by Dan Ariely

This is the book that first turned me on to the fascinating world of behavioral economics. Ariely does an excellent job of explaining many of the core principles of behavioral economics with stories and experiments. Every retailer should read this book to better understand how people (customers) think and behave. It will absolutely open your eyes.

Those are some of my favorites. I’m always looking for a new read. What books fired you up this year?



Retail: Shaken Not Stirred by Kevin Ertell


Home | About