The Rise and Fall of Nate Silver: A Lesson in Risk Communication

Political prognosticator and analytics guru Nate Silver rose to national fame by correctly predicting elections. But in 2016, Silver joined almost every other analyst by projecting a victory for Hillary Clinton over Donald Trump. Was Silver’s good luck over?

Cognitive Bias and the “Failure” of Data

Actually, Silver’s estimate for the 2016 election was closer to correct than almost anyone else’s. He saw Clinton as a heavy favorite, but still gave Donald Trump a roughly one-in-three shot of winning. But the world didn’t remember that part of the projection once the election results came in. They just remembered the part Silver got wrong. Nobel Prize winner Daniel Kahneman has an explanation: cognitive bias.

Kahneman studied how people make decisions and judgments, and he quickly discovered that they don’t make any sense. People like to think of themselves as logical and rational, but they mostly use logic to justify believing whatever they want to believe anyway. And one thing people absolutely love to believe is that the future is certain. Human minds loathe uncertainty. Uncertainty breeds anxiety and fear—sometimes paralyzing fear. So when given a number like “one in three” or “ninety percent,” they subconsciously convert the odds to “yes” or “no.”

This cognitive bias is often very useful. You probably never consider the statistical chance that you’ll be run over by a bus because if you did, you might never leave the house. It’s far easier, and probably mentally healthier, to treat the risk of bus accidents as a 0. But the tendency to round probabilities up or down can be disastrous in the business world.

Communicating Risk

Have you told your boss that there’s a 90% chance you’ll make the sale? If the deal didn’t go through, you were probably in a bit of hot water. Has a supplier ever told you her product’s failure rate was less than 1%? You’d probably be pretty mad if your order was a dud. The problem with both of those statements of probability is that they do a poor job of communicating risk. They invite the mind’s cognitive bias to take over and convert the estimate into a certainty. When that certainty turns out not to be so certain, it feels like a broken promise.

That’s why the world decided Nate Silver was wrong. They had rounded up the probability of a Clinton victory to a guarantee. When Trump won, it felt like Silver had broken his word. His failure wasn’t in the data—it was in the way he communicated the risk.

The lesson here is that quoting numbers won’t save you.

Don’t just toss out percentages—put them in context. Visualizations are one useful technique. If a product will fail one time in a hundred, a graphic with 99 white shapes and one black shape gets the message across far more effectively than the numbers. Analogies are also effective. A 90% probability? That’s about the same as the chance that an NFL kicker will make a 32-yard field goal. Anchoring the numbers to a familiar context creates a lasting impression. It forces the mind to acknowledge uncertainty.

In business and life, people care about honesty. But if your goal is to be trustworthy, it’s not enough to state the facts. You have to make those facts sink into others’ minds. When it comes to probabilities and risks, that task is taller than it looks.

Using Customer Data to Create a More Powerful Customer Experience

Make no mistake: the data revolution is upon us and has, perhaps, affected no industry more than the wonderful world of marketing. Your average marketer now has access to huge volumes of information about who their customers are, how they’re behaving, what their thoughts are on certain issues and more – all of which can quickly prove overwhelming. It’s important to look at the big data phenomenon for what it is, however: an opportunity. It’s a very real chance that marketers have to realign their efforts and create the type of powerful customer experience that creates a loyal army of brand advocates.

Broadening Your Customer Personas

Customer personas have long been a tool marketers have used when trying to relate to their target audience. These fictionalized, typically generalized versions of theoretical people can be a great way to help the designers of a campaign keep their “eyes on the prize,” so to speak. After all, if you’re setting out on a road trip across the country, it can be helpful to know exactly where you’re going before you back out of the driveway.

However, the huge influx of data that marketers now have access to is a terrific way to deepen these customer personas more than ever before. You no longer just have things like age, gender, employment status or income level to work with. You can now draw from not only what has influenced past purchasing decisions, but WHO. You have volumes of analytical data pertaining to lifestyle, interests, and behavioral patterns. You can even draw valuable information from how a person might respond emotionally to a certain event in their life.

All of this means that an already powerful tool, customer personas, can now be put to even more meaningful use in the future. These personas are no longer generalized at all, which is very much a good thing for marketers everywhere.

Redefining the “High Value” Customer

Another great way to use customer data to create a more powerful customer experience is to reassess your “best” or “highest value” customers through the lens of this new data you’re working from. You’ve always been able to call up data like average purchase size, lifetime value, and acquisition costs pretty easily, but now you can go deeper. You can get a real sense of how satisfied your customers are with your products or services and look at how that information may affect what you need to do for your customers in order to get them to remain loyal.

You can also see whether or not the people you’re actually targeting with your marketing materials are the ones who are actually spending money on what you have to offer. If there is a discrepancy there, who ARE your buyers? Is this a problem, or is this a happy accident? What does this new information say about decisions that you were previously making on assumptions? This is all incredibly valuable information to have moving forward.

At the end of the day, the huge volumes of customer data that marketers now have access to is absolutely NOT a burden. We live in an age where it’s now easier than ever to glean the type of valuable, actionable insight that you can use to make more effective, strategic decisions. All of this allows you to drive home the most important benefit of all: creating a much more powerful, organic, and deeply rooted customer experience than what was possible even five short years ago.

Predictive Analytics: One of the Keys to Direct Mail Marketing Success in 2015 and Beyond

Direct mail marketing is still one of the best and most efficient ways to connect to your target audience, even in this social-media-centric world. But that doesn’t mean you need to eschew technology altogether. Case in point: predictive analytics are quickly becoming not just a recommendation, but a requirement for anyone running a direct mail campaign.

What Are Predictive Analytics?

At their core, predictive analytics leverage statistics, data mining, and similar techniques to create a prediction about future behaviors. The idea is to take the past behavior of your target audience and use it to make educated guesses about future activities.

The concept is used in Internet advertising on a daily basis. Have you ever wondered why you suddenly see advertisements for home audio and video equipment or Blu-ray movies right after you purchase a high-definition television set online? It’s a combination of programmatic advertising and predictive analytics at play. Marketers know that based on your purchase, there are certain types of accessories you can definitely use.

If you just bought an HDTV, it goes without saying that you could probably use some shiny new Blu-rays to play on it. By targeting you with advertisements based on that information, businesses know they have a much better chance of making a sale than if they randomly targeted 10,000 people, many of whom might not have an HDTV at all.

Many businesses don’t realize this same idea can also play a very important role in how their direct mail marketing campaigns are conducted.

How Do Predictive Analytics Help in Direct Mail Marketing?

The major benefit predictive analytics brings to the world of direct mail marketing is one of precision. You no longer have to spend time and money each month to send mailers out to all 3,500 people who live in a particular ZIP code. The fact you were sending out materials to many people who ultimately had no interest in your products or services was always just an accepted “cost of doing business,” but that doesn’t have to be the case any longer.

Thanks to predictive analytics, you now have a better chance of targeting the RIGHT people within a particular ZIP code based on their past interests and behaviors. Instead of sending out 3,500 mailers and achieving a 20% conversion rate, you can save time and money by only sending out 1,000 mailers while achieving an 80% success rate at the same time. It’s about giving you a much smarter way to spend your marketing dollars. It’s also about empowering you to stretch your marketing campaign’s strength even further.

In direct mail marketing, success doesn’t mean spending as much money as possible. Instead, true success and market penetration are achieved by spending every dollar the right way. Whether you have $10 to spend or $10,000,000, that theory will always hold true. By making excellent use of advancements like predictive analytics, you can make sure your important materials are actually getting in front of people who find them valuable. This will go a long way toward increasing not only the efficiency of your campaign, but also its general return on investment.