Will Bloomberg Disclose How Heavily Reporters Mined Customer Data? (It Watches Them, Too.)

Let me start by saying up front that I used to work at Bloomberg News, so what I’m about to say is informed by that experience. I worked at Bloomberg for about a year after the company bought BusinessWeek magazine from McGraw-Hill and turned it into Bloomberg Businessweek.

In that capacity, I learned to use the Bloomberg terminal that sits on some 315,000 desks in the financial industry and that makes the company all its money. And I learned early on that practically every move terminal users make is tracked and recorded. (Also, it goes without saying but I’ll say it anyway, this website is owned by News Corp., which owns Dow Jones, which is a competitor to Bloomberg.)

In a Sunday editorial on Bloomberg View, the company’s equivalent of an Op/Ed page, Editor-in-Chief Matt Winkler wrote, “Our reporters should not have access to any data considered proprietary. I am sorry they did. The error is inexcusable.” He opens the piece by quoting from a section of his book, “The Bloomberg Way”:

“The appearance of impropriety can be as damaging to a reputation as doing something improper. Because we hold others accountable for disclosure, we expect the same of ourselves. While disclosing errors of judgment may be embarrassing, the sooner the lapses are reported, the sooner there is nothing more to say.”

This raises the question: How fully will Bloomberg News disclose what it admits to be an “error of judgment” that is “almost as old as Bloomberg News.” How many reporters used the Z function — a software command that displays whether or not a customer is logged in and which functions he or she has been using the most — over the many years it was available to them? Chances are that Bloomberg has the data on precisely how often it was used and by which reporters. It could with some effort call in a third party to perform a detailed audit on this, and then disclose the findings of that audit to clients and the rest of the world.

I’ve asked Bloomberg about this. Spokeswoman Lauren Meller didn’t have an immediate answer. If I get one I’ll post it here.

If you’re going to properly understand the controversy that has emerged about the company in recent days, you need to understand the basics of the terminal itself. Bloomberg is at its very heart a financial data software company. In executing a “function” on its terminals, which are seen as status symbols of the financial industry, you type a command, usually one to four letters, and hit the Go key, which replaces the Return key on the conventional keyboard. When looking up, say, the price and fundamentals of Apple shares, you type AAPL, hit a key labeled Equity to indicate the first four letters are intended to indicate a stock ticker symbol, and then hit Go.

During the year I worked there I never heard about the so-called “Z function” at the heart of the current controversy, but its existence isn’t surprising. If you haven’t been paying attention, here’s what it’s all about. All 2,000-odd reporters at Bloomberg News have these terminals on their desks and use them to conduct research, report, write and publish their stories, and to communicate within the organization and without.

The Z function, now disabled for newsroom employees, showed when clients were and were not logged in to the system. When someone hadn’t logged in in a while, an attentive reporter might see that as a tip that the person was changing jobs or had left a firm, and then the reporter would start asking questions. Stories about executives moving between firms tend to be popular among terminal clients. It also showed what other Bloomberg functions these clients had been using, but in a non-specific way that wouldn’t show what stock or bond or other matter they might be researching or which news stories they had been reading. Bloomberg reporters are also said to have had access to transcripts of customer service calls clients made seeking help with functions.

Bloomberg is and has always been a “big data” company. The newly fashionable idea that you can learn a great deal and thus improve a software application by analyzing the big mass of data gathered about how it is used and where users run into problems has been been at the core of Bloomberg’s operational philosophy from the beginning.

Employees know from the moment they join the company that the amount of time they spend at their desks is logged. Building security systems are linked to the terminal and an access badge. When you “badge in” at any Bloomberg office around the world, this occurrence is logged. If you’re a Bloomberg employee based in New York and happen to be visiting London or Tokyo, your arrival and departure times are tracked, as is the amount of time your terminal is idle, should you be out gathering news or taking a lunch break.

I never experienced this first hand, but I heard privately shared tales from colleagues about their annual performance reviews, and discussions would at times turn to how well they used the terminal to do their jobs. As I was first joining, a friend who had worked at Bloomberg for a while told me that during one such conversation, he was mildly scolded for using Yahoo Finance to look up some bit of financial data — terminals also have Web browsers — rather than the terminal itself.

I point out this conversation for a reason. In its quest to make its products better — certainly a logical goal — Bloomberg clearly tracks how often its clients use its many functions. If that’s true, then it logically follows that it tracks how its reporters do the same thing. Historical data on the use by reporters of the Z function exists and can be examined.

As an organization, Bloomberg News is journalism as re-imagined by Fredrick Winslow Taylor, the philosophical father of factory automation. Its reporters are routinely gauged on how often they log scoops that appear on the terminal’s news service. Taylor believed that by analyzing work, the “One Best Way” to get it done could be found. As the founding editor of Bloomberg News, Winkler always struck me as an avid student of “Taylorism.” Every bit of data that can be gathered is analyzed to make the news gathering process better and more efficient. Indeed, Winkler’s 360-page book seems almost Taylor-inspired.

Scoops and other distinctive stories are further categorized into a taxonomy using an internal nomenclature, and the best of those are singled out in what’s known as “Matt’s Note,” a weekly memo from Winkler. An MMWin, or a market-moving win, is a story that beats a similar story by a competitor by several minutes and that after publication causes the market to react in some way. A Follow is when a competitor writes a story that follows up on a Bloomberg scoop. And there are many others.

All of these are thought to be tracked and used to evaluate a reporter’s performance every year. That means there’s data on how many reporters used the Z function and about whom, and probably data as well correlating to the published stories that resulted.

With at least two large banks complaining, and now two government entities — the Federal Reserve in the U.S. and the European Central Bank — asking questions about all this, you can expect concerns about the lines between Bloomberg’s business and news operations to persist.