You Can’t Kill the Bloomberg Terminal. But If You Were Going to Try, Here’s How.

“I’m going to kill X” is usually a bad business idea. Generally, there are obvious reasons to kill it (it’s profitable, customers hate it), and there are nonobvious reasons nobody has succeeded yet.

Bloomberg is one of the best software companies in history, and it may be the single best service I’ve ever used — the user interface on the Bloomberg Terminal isn’t great, but the customer service is incredible. If it breaks or doesn’t do something you need, the company is willing to fix it or build it.

When people try to understand the Bloomberg Terminal as a product, what they often miss is that Bloomberg is a great business because it has low churn, and it has low churn because its IM function is basically an exclusive club of approximately 325,000 people whose employers think they are worth at least $25,000 per year above their base salary. Like any exclusive club, if you can get in, it’s assumed that you’re probably worth talking to.

There are some ways to make inroads into Bloomberg’s market. For example, Money.net is a solid rendition. Its founder worked at Bloomberg and tracked oil. (His book was one of the sources for my fracking piece.) Koyfin looks great, offers a wealth of data, and has an absurdly responsive site. A broker might soup up their app or website to peel off some Bloomberg users (maybe offering similar functionality for free to customers who pay high commissions). You might build something for tracking and discussing esoteric cryptocurrencies on Urbit (or a similar system that builds identity and exclusivity directly into the protocol). And the Bloomberg killer might already exist in China, which has a big market, lots of institutions, and lots of retail traders. It’s also willing to protect domestic software companies from multinational competitors. Given the network-effect dynamic, I probably won’t hear about the Bloomberg of China until it’s a threat to the U.S. version, at which point it will already have more users.

Bloomberg’s value prop and the opportunity for competitors

The key point to understand about Bloomberg is that it’s both a software product and a social network. The software product determined who would join the network, but the network is what keeps users there. It’s like a multiplayer video game, or Harvard: Sure, the quests and campus are useful, but people keep showing up because of the friends they’ve made or the connections they intend to make.

“If you’re competing with Bloomberg, you’re in the messaging business, and everything else is designed just to make sure the right people are signed on at all times.”

Bloomberg is not the only finance-related messaging platform that has ever existed, though it’s certainly the best monetized. For a long time, AIM was synonymous with U.S. equities. Literally until the day it was shut down, I knew people whose AIM handle was in their work email signatures. A year after I left a hedge fund and went to work at a cryptocurrency company, I idly signed on to AIM, and within a half-half hour, someone sent me a quick message summarizing a management meeting he’d just hosted with a semiconductor company. Yahoo Messenger is allegedly big in oil trading, and WeChat is essential for Chinese stocks. What drives up the premium for Bloomberg is that it’s for finance and not much else.

The only non-finance content on Bloomberg is news (to read during your downtime), the classified ads section (where you can spend your bonus or sell stuff if your bonus didn’t meet your expectations), and the MVP page of the most-viewed profiles. If you’re competing with Bloomberg, you’re in the messaging business, and everything else is designed just to make sure the right people are signed on at all times.

This looks superficially like an opportunity, because there isn’t a good free finance site. You can say that about a lot of topics, I’m sure: Maybe there isn’t a good modern art site or a good rugby site or whatever. But I’m confident it’s true for finance because there used to be. The quality of free resources for investors has dropped since I first got interested in the topic in the late 1990s.

I haven’t found a free stock screener that’s as good as what MSN had in the late 1990s and early 2000s, nor is there a message board quite as lively as Yahoo Finance. (Stocktwits is better as a user community, but the relative dearth of long-form content means it’s not as good for tracking a single stock.) Value Investors Club is about as good today as it was back then, and the design shows.

There are a few reasons for this decline:

  • Investing as a hobby peaked in the 1990s and has declined since then. Peter Lynch was an incredible stock picker. (Arguably, if you adjust for his factor exposure, Lynch outperformed Warren Buffett by a wide margin.) Buffett is a big celebrity, but he was probably the biggest investing celebrity then, too. Today, more professional investors work for private companies than mutual funds, which has sucked a lot of the oxygen from the room.
  • Related to this, one reason there were so many resources for individual investors in the 1990s and 2000s was that the population who could build a website had a high net worth and lots of public companies they could invest in. The Venn diagram of people with a six-figure E-Trade account, an interest in trading dot-com stocks, and a working knowledge of Perl and HTML was basically a blurry circle. After the IPO market dried up, many of those types were more likely to become angel investors than public equity investors, and indeed, the post-bubble internet has a lot more information on startups and venture capital relative to public stocks.
  • Back before we figured out online media business models, there was more malinvestment. Very few online media properties made much money in the late 1990s, and if they did, it was because they got ads from other internet companies. Once people started tracking ad performance more rigorously, they figured out that extensive media coverage and market data for mid-cap and small-cap stocks just didn’t pay as well as news coverage of big companies, monetized through ads for funds and robo-advisers.

Figuring out product parity

A viable Bloomberg killer has to do everything Bloomberg does, at least for some early users’ definition of “everything.” That’s a tautological claim, but an important one: Since nobody uses the full set of Bloomberg’s tools, everyone has a different idea of what constitutes a replacement. If you’re using Bloomberg to figure out good M&A targets, your idea of what it does is very different from someone who spends all their time using Bloomberg’s options pricing tools or its incredible supply chain maps.

Suppose we start with equities. According to this interview with Morgan Downey, the founder of Money.net (the leading contender for a Bloomberg killer), the backend for getting real-time data from exchanges to customers costs $10 million to $20 million. Then you need fundamental data. (Scraping SEC filings is doable, but there are many edge cases. Companies use all kinds of weird metrics and inconsistent labels in their financial statements, and most people I know still don’t export financials directly from Bloomberg unless they’re grabbing aggregate data for an industry.)

And then you have to figure out news. Not just licensing news from newswires, scraping blogs, and grabbing company press releases. You also have to instantly summarize. When a big news story hits, Bloomberg will run a series of quick all-caps headlines summarizing the key points. What’s “key” varies from one company to another, as well as in context. It’s almost never earnings per share (the metric CNBC highlights), but it’s often some underlying driver of top-line growth. For one company I followed in the past, the important metric wasn’t its revenue, but its revenue guidance. It made a habit of reporting earnings when the quarter was more than half over, so the company’s “guidance” was a guess about 13 weeks’ worth of performance based on eight weeks of data. (Before Bloomberg headlines got good, many basis points of alpha were earned by people who knew to scroll down to the top third of the second page of the release and react to bullet point number three.)

That’s the kind of knowledge you can accumulate either very slowly, through trial and error, or very quickly, because being yelled at by 0.1% of your 325,000 customers still means fielding 325 phone calls representing $8 million in revenue.

That’s one company, and that’s what matters for now. Maybe next year the only thing that matters for that company will be gross margins or customers. Or maybe next earnings release, its CFO will resign to spend more time with his defense attorney. Parsing text documents into relevant headlines in real time for people who will gain or lose millions of dollars from your work is not easy, and even if you get it right 99.99% of the time, people will remember the 0.01%.

But it can be done. We’ve started with up to $20 million in capital costs, and now we’re adding at least that amount in ongoing product development and fine-grained product analysis costs. Maybe it’s time for some customers!

Building the viral loop

The right time to launch your Bloomberg killer is about a year before the next bubble in a liquid alternative asset that retail investors want to talk about and institutions want to trade. In other words, the right time to launch was 2016, so you’d catch the Bitcoin wave.

There are echoes of this in Bloomberg’s early history. When it started, Bloomberg was a platform for bond market data. Your natural attitude toward bonds will depend entirely on whether you first learned about finance as a retail investor or as a professional. To retail investors (that’s where I started), bonds are intrinsically boring; they are, in finance theory terms, required to be less interesting than stocks, since a corporate bond has less of the downside and less of the upside. By that theory, government bonds are even more boring: A country’s tax base is basically a diversified index fund of all its taxable economic activity, which is even safer.

If you’re a professional, you know this bias is totally wrong: Bonds can be more fun than stocks, because given enough leverage, you can make a portfolio of treasuries just as wild as a concentrated bet on penny stocks. Since the finance industry is quite levered, participants tend to understand this.

So, starting with bonds was the right move for Bloomberg. It was also the right move for Mike Bloomberg, whose previous job was at bond trading powerhouse Salomon Brothers.

Bloomberg’s early history matches the advice Y Combinator gives new startups: It’s better to be essential to a small number of people than to be nice-to-have for a lot more people. Soon after it launched, Bloomberg was the single best place to get information on bonds, and it was, naturally, a good venue for trading them as well.

Bloomberg became a necessity for bond traders and then moved into other assets. It started out as more of a raw data source but quickly branched out into news. The company knew news was a big deal soon after launching, when Bloomberg was the first outlet to break the news that Jerry Garcia had died, which crashed its servers.

Bloomberg’s retention is a function of its existing user base. If all analysts, fund managers, and salespeople are on Bloomberg, none of them can afford not to be. Social networks are tricky because network effects work, but they also work in reverse. If the most valuable members have a reason to stay (for example, top salespeople and recruiters on LinkedIn, media personalities on Twitter, and open-source contributors on GitHub), the business is pretty stable.

A social media site is easy to compete with if the most desirable users tend to leave. Dating sites, for example, are unstable because the most attractive prospects pair off, lowering the quality of the remaining user base until the site can’t recruit new users. The alternative is to keep users constantly hooking up. In other words, every dating site is either turning into a dying clone of eHarmony or a cheaper clone of Backpage.

“One option for bootstrapping a Bloomberg clone is simple: Buy back AIM’s source code and create a multiplatform chat app.”

Bloomberg started out as indispensable for treasury bond traders, so it became indispensable for people who traded other bonds, and then for people who traded equities, and then for people who traded anything, even entire companies. And now it’s quite sticky. Any meaningful drop in overall asset prices or the share of actively managed assets will surely shake some people out, but they’ll be the most marginal users anyway. When people downgrade from Bloomberg, I’m sure it’s painful, but only upgrades would be fatal.

There isn’t any asset class I would describe as “the next U.S. Treasury.” (No, not Chinese government bonds; they’re more like the next 1980s LatAm bank loan.) The next Bloomberg will have to start smaller, because the first Bloomberg ate all the low-hanging fruit. Markets aren’t static, however, and it’s possible to build a viable user base focusing on some narrow product and extending from there.

Alternatively, you can buy your initial user base.

One option for bootstrapping a Bloomberg clone is simple: Buy back AIM’s source code and create a multiplatform chat app. Your user base won’t be huge, but they will mostly know each other, so your app won’t quite start from zero. This was a more plausible plan before AIM was shut down in late 2017. (Until the end, I was hoping Bloomberg would buy it.) But the source code probably still exists somewhere, and it’s not like it’s being used for anything else right now.

Full-screen time retention: The one true metric

Suppose you’ve built your basic product, and you have your initial users. What do you care about next?

A few years ago, I talked to a small startup that was hoping to displace Bloomberg. It had done the usual back-of-the-envelope math: You can copy 99% of the functionality you use for cheap, charge 95% less, capture a bigger market, and eventually make inroads on institutional investors. Every time the market slumps and costs get cut, your corporate market share rises; meanwhile, as long as you don’t mess anything up, your retail share stays high. And you can further benefit from executive distraction, since Bloomberg’s eponymous CEO has some big plans involving government work.

This particular startup didn’t work out, but it was a fruitful meeting because it clarified why this kind of startup is so hard. Bloomberg works because it minimizes churn—that is, it minimizes the odds that a user will ever stop using it. And it’s almost tautological that the way to make sure someone keeps using your product is to ensure that they never stop using it.

Hence my One True Metric: A Bloomberg competitor will win only if users keep it full-screen on at least one monitor for the entire workday. (Building a mobile app is a given; you need to ensure that you don’t miss crucial minutes of usage during lunchtime or bathroom breaks.)

If you visit a trading floor, you’ll see that Bloomberg has pulled this off: Between the portfolio monitor screen, the IM window, the news ticker, and various data screens, it probably has more pixel share than any product besides maybe Excel. (And the ratio of Bloomberg time to Excel time is a sort of status symbol in finance.)

If you look at the way retail investors operate, you’ll see there isn’t anything that quite matches Bloomberg’s ubiquity. A few finance sites have some combination of features and search engine distribution—if you search for a ticker symbol, you’ll get results powered by that search engine’s finance property—but nobody has welded together financial data, real-time quotes, and market commentary in a useful way.

If you wanted to make this happen, you’d have to start out painfully small. Take some subset of the market that has a combination of:

  1. Lots of retail interest.
  2. Lots of news flow.
  3. Low but rising institutional interest, such that tracking it isn’t high on Bloomberg’s priority list.

At a minimum, you’d need real-time prices and real-time chat. Then you’d need aggressive moderation — many-to-many communication plus retail investors equals spam and fraud. If you got this up to speed in 2017, you’d have the world’s best cryptocurrency news and data site, with a couple hundred daily active users who logged 10-plus hours a day. If you hit your stride a year later, it would be the same thing for legal marijuana stocks. (I have no idea what will catch the median retail investor’s fancy in 2019.) I’m not sure what the next opportunity will be. Bloomberg but for YieldStreet’s alternative assets? Bloomberg for buying influencer engagement futures? Bloomberg for prediction markets?

Not only is the exact market hard to predict—the size is unpredictable as well. Maybe the next fad will be a cultural phenomenon like Bitcoin was (briefly). Or maybe it’ll be like MoviePass, which was also a cultural phenomenon but produced orders of magnitude less market capitalization.

Whatever it is, it will feel depressing, because “success” will look small. But that’s because a lock on a tiny market is the only viable place to start. If you copy most of what the Bloomberg product does, you just have an almost-as-good app without the network effects, so you’ll be fighting an endless losing battle against churn.

But if you can be 200% as good as Bloomberg for just one asset class, you can do it again, and if you pick your asset classes right, you’ll have a market that grows with you. (This dynamic shows up in every bubble: Once there are success stories, other companies will pivot into being similar enough to a hot market to get a higher valuation. Lots of peer-to-peer ideas from the early 2010s finally got funding as cryptocurrency projects in 2016 and 2017.)

The painful reality, though, is the contrapositive of network effects. Easy growth at scale means a surprisingly long and painful slog to achieve scale. Airbnb had to launch its product three separate times before it got traction, and only its low burn rate and four sigmas of scrappiness saved the company.

If you try to raise a Series B because you’ve topped out at under 1,000 daily active users, and yours is the cheap version of a product that has fewer than 325,000 users, you’ll need some incredibly farsighted venture capitalists. At every fundraising stage, the right move for a Bloomberg killer is to trade better user-level metrics for higher user counts — prefer higher time spent per user to higher daily actives, prefer daily actives to weekly and monthly, and prefer live users to account registrations. But each of these metrics means lower immediate revenue and less flexibility in exchange for a longer runway.

Suppose your Bloomberg killer survives long enough to raise a B or C round. It’s the go-to source for tracking and trading a few financial instruments — maybe cryptocurrencies, small-growth stocks that don’t have sell-side coverage, and obscure fixed-income securities that don’t have CUSIP numbers. At that point and every point along the way, the temptation will be to monetize more by adding low-quality users. But low-quality users dilute the experience for the engaged users by reducing the information density of the app and the utility of in-app messaging.

The two traps are appealing to every institution and every retail investor. If you appeal to every retail investor, you get an unusably spammy platform; if you try to appeal to every institution, you get a product that’s a sad consolation prize for people who don’t merit a Bloomberg. The only way to win is to have a user base, however tiny, that tries out Bloomberg and says it’s simply inadequate for what they need to do.

It’s implausible that even the most farsighted founders and venture capitalists would consistently turn down a more profitable, lower-risk strategy while dumping ever more capital into the high-risk alternative. A viable Bloomberg competitor isn’t fundable, and a fundable one isn’t viable. So, for the foreseeable future, I view Bloomberg as unkillable.