Moneyball, Big Data, Analytics, Correlation, and the Evolving World of Marketing

Recently, one theme that's been running through much of my thoughts is wrapped up in the ideas presented in Moneyball, the book and, more recently, movie starring Brad Pitt. And while I have not seen the movie or read the book (other than a short excerpt), I've listened to several extended interviews with Michael Lewis.

Moneyball is sort of about baseball, but what makes it interesting for marketing is the analytics theme. You're probably already familiar with the basic story -- A's general manager Billy Beane takes over and is forced into an extremely low payroll. In order to be competitive in a league where the New York Yankees can afford to spend more than three times his budget, Beane turns to analytics and statisics to look for undervalued players and player characteristics that can help win games on a budget.

There are a lot of interesting posts on Moneyball and business. Here's a good one, Moneyball – Lessons from Baseball for Voice of the Customer, that I thought pulled together a good summary. If you had to put it all together, you could boil it down into a core recipe:

• Measure as much as possible
• Log your data
• Look for correlations (or non-correlations) -- find the numbers that matter
• Question conventional wisdom
• Learn to play by the numbers

Keeping stats isn't new. Baseball tracked batting average for years, and players were often listed by batting average. But the team with the best batting averages didn't always win. In the early days of the web, the big measure was clicks. If someone convinced you to post an ad on their site, they would then tell you how many clicks the ad had -- and that was considered the measure for success. Like tradeshow badge scanning leads, these numbers could easily be padded with tricks like cool give-aways. Clicks don't equal quality.

Modern marketing runs on web scale. With today's web, we interact constantly with the data engine, supplying test results that dwarf some of the some of the most sophisticated focus-group programs of the past. With tools like Google's Website Optimizer and A/B or Multivariate testing, even small businesses with modest marketing budgets have access to sophisticated experiment engines. But, just in case you're one of those old-school marketers and you missed the memo, a revolution has taken place and everything has changed.

Analytics versus Design
In the web world, we've been working with analytics for quite some time. One of my favorite stories about the power of analytics comes from a time when I was visiting Google a couple of years ago. There, they were showing off their Web Site Optimizer tool and talking about how they had looked at redesigning the page to look more like their Google Analytics page. However, when they ran the A/B and multivariate tests, they found that the new design didn't perform as well as the existing design.

Contrast that with this quote from a post about Fab on PandoDaily:

The difference was pronounced in a recent meeting Goldberg had with a Valley-based recruit for a technical position. Within in ten minutes of the interview the two were fighting. Goldberg asked what he’d do with the Fab homepage, and the recruit gave the usual spiel about A/B testing the layout to see which products made people click more, and how the data said they should be laid out on the page. He called the product placements on the front page “ads,” and Goldberg balked. They aren’t ads, he said, they’re editorial. “We aren’t trying to make people buy certain things, we want to guide them through a story,” he says.

So which is more important, Design or Analytics? Style or Stats? Perhaps, more importantly, can you find the right balance between the two?