Companies have been told they should beg, borrow and steal an enterprise data strategy from outside their industries to jumpstart their own.
At the 2013 MIT Sloan CFO Summit there were plenty of data strategies to ogle -- from Wal-Mart Stores Inc. to LinkedIn Corp. to the big daddy of big data, Google Inc. But economist and summit panelist Alberto Cavallo took a different tack.
The assistant professor at the MIT Sloan School of Management urged chief financial officers (CFOs) to start cultivating talent in their own backyards -- especially if their companies hope to reap business benefits from data. Why? Because your own people are the ones best equipped to know a data breakthrough when they see it. Correlation does not imply causation. As Cavallo put it, Argentina's rise in inflation isn't brought on by solar flares, which both happen every 10 years. "I know Argentina is a difficult country to understand -- I come from Argentina," he said. "But I can assure you, there's no relationship between those two things, even though they are, in fact, statistically correlated."
In other words, investing in relevant data skills does not necessarily mean buying data scientists. "Up to now, this has been mostly a computer science revolution," Cavallo said. "But what's often missing is people from other fields who can [consider] these new data sets in a way that's going to let me ask specific questions important to my business." Building an enterprise data strategy is an inside job.
Of course, Cavallo has access to the computational prowess of MIT, so data science skills aren't hard to come by. But he also said this: "Some of the best people you bring on board are people who have no data skills." They can see the big picture and ask the right questions because they aren't tainted by preconceptions of what the data should say or how it should look.
Oh yes, and start small. And by small, Cavallo means start with a small data sample. The big challenge with data today isn't size, it's variability, he said. So, starting small gives practitioners a chance to learn about the data and figure out how it can be combined with new data types. From there, they can quickly scale up the amount of data they're working with, leaning on technologies such as cloud computing to do so.
The finance of algorithms
Finance and marketing are getting a little cozier -- at least at Wal-Mart. Algorithms built for search engine marketing -- online auctions that bid out advertisement space based on what a user types into the search box -- require financial guidelines.
"The marketing team would love to spend as much money as they want and get as many customers as they can," Liz Coddington, vice president of finance for Walmart.com, said. "But we have the financial constraints of the business that we have to manage."
Previously on The Data Mill
Crowd sourcing is the new cloud computing -- get with it CIOs
Cool or creepy? The ethics of big data is on the table
Relational databases are far from dead -- just ask
She has to understand how the algorithm works on a pretty detailed level so she and her team can set those constraints for any bid made. And it also means any tweaks to the algorithm are run by Coddington directly for approval, she said.
At Dunkin' Brands Group Inc., the marketing and finance departments are also becoming more entwined. There, the business analytics team sits within the finance department and reports directly to Paul Carbone, the CFO. Who is the team's number one customer? That would be the marketing department.
"[The team] probably gets 80% of their direction from the chief marketing officer, and 20% from me," Carbone said. The split provides some good ole "creative tension," he said, but, in fact, the organizational structure more closely resembles the model of governance known as separation of power.
"We consider ourselves the seekers of the truth," he said, "so we like to believe we're a little less biased."
CFOs are also getting social. Coddington talked about how Walmart.com uses social media data to pinpoint trends before they are, well, trends. "We want to make sure that Wal-Mart -- both on the website and in the stores -- has the products customers are looking for," she said.
But social networks don't exist solely on platforms like Twitter or Facebook. There are also "implicit social networks," said Alfred Spector, vice president of research and special initiatives for Google. When customers purchase something from Walmart.com or Amazon or any number of retailers, they share a natural link with other customers who bought the same thing. Amazon, for one, has mastered how to use those connections to suggest additional items customers might also like.
"That's called collaborative filtering," Spector said. "And that's an extremely important implicit kind of social network analysis."
"The greatest change here is not necessarily the size of data, but the new types of data that are available to us." -- Alberto Cavallo, assistant professor, MIT Sloan School of Management
"Innovation without failure doesn't exist." -- Bill Aulet, managing director, Martin Trust Center for MIT Entrepreneurship
"One comment I'll made on talent, and if I say this, it might not get me invited back next year, but at State Street this year we've hired more engineers than MBAs." -- John (Jack) Klinck, executive vice president and head of global strategy and new ventures, State Street Corp.
"Innovation is a very significant part of what Fidelity does. It's supported both financially and by other means, and it's a wide set of activities that each of the individual business units are encouraged to pursue. As it happens, for me, I'm involved in an effort to monitor what results we are getting from that investment spending." -- Alan Scheuer, CFO, Fidelity Investments
"If you want to store the raw data of all of the text in the Library of Congress using commercial prices in the Google cloud services that we provide to the general public, it's around $5,000 a month -- with some margin of error." -- Alfred Spector, vice president of research and special initiatives, Google
This was first published in November 2013