In 2001, when Dan Themig co-founded Packers Plus Energy Services Inc., a Calgary-based company that creates equipment used for fracking in the oil and gas industry, most of what he did was based on a gut feel and crude data. So when he asked his manufacturing manger how much product he would need for an upcoming year, he usually didn’t question his numbers.
In 2008, though, Mr. Themig began using predictive analytics to help determine the demand for his services. After looking at about 14 different data points, such as natural gas prices, oil rig counts, weather patterns and more, he saw that demand for the coming year was going to spike dramatically. He asked his manufacturing manager how many more packers – a tool used for fracking – he’d need for the coming year and was told about 25 per cent more.
“I said we needed double,” Mr. Themig says. “He went dead silent and then asked, ‘How do you know that?’ ” Making twice as many products was no easy task. The manager had to purchase a large percentage of the steel in North America to get the job done.
In the end, his data didn’t lie. He produced, and sold, exactly what the numbers told him he would.
Predictive analytics – using data to forecast everything from future sales of a product to when a piece of equipment might break down – has come a long way since direct mail companies in the 1970s first used numbers to determine whom to send pamphlets to. Back then, data was used to predict buying habits; it’s now used across myriad sectors, such as oil and gas, mining, health care and even sports.
Richard Boire, a partner at Pickering, Ont.-based Boire Filler Group, says that on a macro level, predictive analytics is used the same way in every industry. “It’s the use of historical information to predict a future event or outcome,” he says. Companies look at patterns relevant to their business and then use that data to become more efficient operations.
When you look at what individual sectors are doing, though, it becomes clear that predictive analytics help with almost anything. Mr. Themig uses data to determine when the machines that make his fracking tools will break down and whether or not it makes sense to replace them. Baseball teams look at historical data to predict a player’s future performance – read the book Moneyball, says Mr. Boire – while financial institutions look at how people spend and save to determine whether or not someone will open, say, an RRSP account in February.
It’s likely more companies and sectors will adopt predictive analytic tools going forward, says Murat Kristal, an associate professor of operations management and information systems at York University’s Schulich School of Business. Today, programs can handle massive amounts of data and spit out results in seconds, he says. Analytics programs are also becoming cheaper to buy.
The only barrier to adoption, and this applies across sectors, is that companies need people who understand how to analyze the data. “The resource constraint is talent,” Dr. Kristal says. “They’re scarce.”
CEOs can look at data themselves, but without someone who can see how each number relates – and often one data point has an impact on another – the analytics are useless. “People need to understand what they’re doing and assumptions behind the analysis,” he says.
Mr. Themig has someone who can crunch the numbers, but all the heads of his divisions are trained on how to use analytics, too. He’s got a lot to pay attention to. He looks at drilling activity in specific areas, energy demand in Russia and China, commodity prices, and much more. He’s currently looking at all of this data to help him predict drilling activity in January, 2014. If he can determine that, then he’ll know how much steel he should buy today.
He also uses predictive analytics to help him suss out bad advice. In 2010 a senior economist visited the company and said natural gas would be $6.50 by 2013. Using that price, one would assume that there would be about 2,500 oil and gas rigs in the United States.
Mr. Themig’s models showed that gas wouldn’t rise to that number, that it would be about $4. In that case, the rig count in the U.S. would be closer to 1,600. Mr. Themig’s data was correct. “Having good data allowed us to have an understanding that what the economist was saying wasn’t going to happen,” he says.
The continuing goal at the company, he says, is to reduce costs and run a more efficient company. Predictive analytics has helped him do that. It’s allowed him to better control his inventory and deliver products at the right time, and it’s also helped him understand where his business is headed in the future. “I’ve been at a lot of companies in my career that didn’t do a good job of predicting where their business will go,” he says. “One predictive item needs to be what technology should we be working on? By using analytics, we can see what the customer will be doing before they even know they should be doing it.”
While he says all his analytics definitely pays off, there’s one computer system that trumps all the rest: the human mind – and gut. It’s ultimately up to him to determine what numbers to pay attention to and whether the data agrees with his gut feelings.
“The entrepreneurial instinct is still the ultimate predictive analytics tool,” he says. “But behind it, you can have sophisticated data and an understanding of how analytics work.”