Very early in the birth of data mining, some statisticians went to a catalog company that was trying to up-sell. The company had been making purchase suggestions to customers based on past experience and intuition: If someone bought a pair of brown slacks, offer them a brown belt.
The statisticians offered a deal too good to pass up: Offer customers what we tell you to offer them, and we won't charge you unless you make a profit. After mining the data, the statisticians told the company: "When the customer orders the brown pants, offer them a camping ax."
And it worked. It's an important story because researchers at MIT have discovered that companies that get the most competitive advantage from analytics use data-driven insights - even when those insights challenge commonly held beliefs of senior management. Can you imagine how some people at the catalog company reacted to the brown pants/camping ax suggestion? Yet they accepted it.
But to get to the point of acceptance at most companies, you need to have laid the groundwork that allows employees to think: "Yeah, let's try that and see what happens.''
There are four things you need to do to make sure your company doesn't just accept the nonintuitive, but embraces it.
Look Beyond the Technology
The technology to use the data is critical - as is the data itself - but it is equally critical to have the right resources in place. It's not just about hiring people with specific skills. It is about hiring people with the aptitude to learn. New sources of data and new techniques for managing data are coming at us all the time, so the skills you hire today might not be the skills you need tomorrow. This is why your hiring should be based on acumen more than skills. You need people who are at ease with math, and who know how to model a problem.
And you need people from diverse backgrounds - nimble thinkers who look at a problem in different ways and bring their experience to bear on it. This is why I look for people who have more than just coding skills. I look for people who have interests beyond their career, who give back to causes beyond themselves, and who have encountered obstacles in life and managed to overcome. They're the kind who will be resilient enough to adapt as the industry continues its fast-paced change. Bottom line: Change the way you think about the people you're hiring. Take a fresh look at your recruiting practices and make sure they're future-proof.
Build a Collaborative Culture
Once you have your people in place, you have to get them to play well with others. So the next item on the checklist is to ensure collaboration between IT and the business. It should go without saying that agreeing on shared goals is essential for success. But business and IT people tend to speak two different languages. People from statistics and engineering backgrounds might lack the strategic view required in the business world, while people with business brains might lack the technical and analytical know-how. And sometimes the disconnect happens because the goals of business and IT are at odds. Whatever the reason, it's important to clear these roadblocks before you start your analytics project. The right mindset to have? Analytics is a corporate service that's sponsored by the business and enabled by IT. Build bridges between functions, because you need each other to succeed.
Keep an Open Mind
How do you get the CMO to agree to market camping axes to brown pants shoppers? By using the language of margins. In competitive fields, every small edge counts. A company we work with told us a 1 percent gain in accuracy would add $100 million a year to its bottom line. Small percentage gains can make a huge difference to your business.
Another way to counter preconceived notions is by devoting time to organizational change. Accept the fact that getting the technology right is only half the battle, and then go to work to sell your colleagues on the value of data-based decision making. Nominate a champion, publicize successes, and share results. Mentally prepare yourself that this won't necessarily be easy. By definition, using analytics to make decisions means you're challenging the status quo. But the rewards are great.
Invest in Training
And finally, embracing data-driven decision making means investing in building and training a team. You need people who understand the data and who have a specialty in managing it. You need someone who understands the math. And you need someone who understands how to communicate results and turn them into actions, because knowing something and not being able to communicate it makes it worthless. While hiring the right people is important, it is rare to get someone who can do all three. We sometimes call these people "purple squirrels," and if you happen to find them, by all means, retain them with stacks of gold. But most likely, you're going to have to build a team through high-quality training.
Many people think that the biggest companies somehow have analytics all figured out, but that's not what we see. In our view, companies of all sizes are still feeling their way through the process of using data to succeed and grow. One thing is certain: It's not optional. The first bank that came out with an ATM gained substantial market share. The second gained a smaller amount. The third added ATMs just to stay in business. Analytics is like that. You're either figuring out how to get commercial value from it, or you're losing. Winning with analytics takes training, and it takes building a culture. It doesn't just happen naturally.