Like bacteria, big data are lurking in the stomachs of cows. Some farmers are using sensors and software to analyze it and predict when a cow is getting ill.
Just like customers, cows do not always speak out when something is wrong. But companies can use data to predict potential risks and opportunities in cows and customers alike.
The message of a new book, "Big Data @Work," by Thomas H. Davenport, a fellow of the MIT Center for Digital Business, is that companies are only beginning to understand the questions they can ask of their vast stores of data — and how to build the internal structures to make the most of it.
"Big data" is a fashionable, sometimes overused term for the vast amounts of information that can now be stored because of the growth of online activity and the low cost of storage.
While companies are busy talking the talk and hoarding the information, according to studies cited by Davenport, only a tiny percentage is analyzed in any way and by only 28% of companies.
In the book published by Harvard Business Review Press, the author has written a guide for those leaders still puzzling about how this fashion fits their business. He insists that companies think up questions to ask the data rather than just playing with it in the hope that magic will happen.
Previously, companies used data to help solve problems, but now they need to develop the "always on" capability to search out opportunities, he argues.
The book is at its best when offering examples that could spark ideas.
The casino company Caesars Entertainment uses data to spot when gamblers have lost so many times at the slot machines that they might not come back: "If the company can present, say, a free meal coupon to such customers while they're still at the slot machine, they are much more likely to return to the casino later."
That example may sound somewhat cynical. But elsewhere Davenport notes how London's Heathrow airport increased the number of on-time flights from 65% to 80% in just two months after using an algorithm to coordinate everything that goes into a flight turnaround process.
Telecom company Verizon has a unit that analyzes location data for other businesses — for example, telling a basketball team where the fans at their stadium came from.
But Davenport underplays these examples, running them alongside case studies of managers stressing the (quite obvious) importance of return on investment.
He focuses a tad too little on enticing examples of big data's uses, such as the farmers who analyze the information in their cows' stomachs. He also has a penchant for "idiot's guide"-style checklists.
More importantly, Davenport shies away from the frontiers of big data. For example, he skates over its use in workplace monitoring to understand employee productivity.
The book fantasizes about a future in which facial recognition can spot poorly behaved dogs in a pet food shop, but neglects to mention the ways that facial recognition is already being used by some companies.
"Big Data @Work: Dispelling the Myths, Uncovering the Opportunities" is full of advice on the kind of technologies used in big data analysis and how to get the right workforce — a real problem because universities are only just starting to create big data courses.
But for a "how to" guide that itself scorns "techno speak," it is surprisingly heavy on the jargon.
It is not clear what there is to be learned from paragraph upon paragraph about which companies call their head of big data "chief digital officer" or "chief analysis officer" or "vice president of customer optimization and data."
For managers unsure what to do with the data piling up in their vaults, the book is worth sticking with. By the end, the cow's stomach may be healthy but the reader's stomach is heavy with the weight of the big data equivalent of porridge.