Watching 60 Minutes this weekend, I was struck again by the old “algorithm” myth that pervades the tech community. As evidenced by Google’s IPO in 2004, tech investors are transfixed by “the algorithm.” Indeed, the tendency goes back even further.
During the first dot com boom, companies with real revenue would often spin off any part of their business that reeked of “services,” with the logic that the services businesses had set values, with a revenue multiplier of 1-3 times. That is, if a business offered people services, the valuations would be dampened. (Wired was a classic example of this, selling off the magazine, which had real revenue, with the aim of achieving a higher valuation for its less profitable digital arm.) A law firm that makes $10 million a year would consider itself lucky to sell for $30 million. This is a near universal rule, applying not only to law firms, but also ad agencies, professional services firms such as Ernst & Young (an old employer of mine), and consulting companies such as the vaunted McKinsey. During the boom, companies would ditch their services arms and rebrand themselves as a tech company, allowing for higher multiples.
With tech companies, the sky’s the limit. The reason that Google could achieve a market cap of $23 billion from its IPO—because it was an “algorithm.” Because the machines could just churn and make money, sorta like a quant. The same logic has applied to the valuations of innumerable tech companies since. Twitter. Facebook. Foursquare. Tumblr.
Now, it would seem to me that a good measure of a company having some sort of algorithm that just churns out money—and can infinitely scale—would be revenue per employee per month. If they’re just printing money, it ought to be like 15 employees sitting in a room making something like half-a-billion a year each, no? They should at least be, you know, making more revenue per employee than, say, McKinsey. And anyway, there’s really no other metric to use. We could use profitability, of course, but Groupon isn’t even profitable yet.
The thing about this that irks me, though, is that many of these companies make their money off of advertising. Advertising doesn’t sell itself. Originally, Google AdWords was primarily a self-serve product. But over time, a sales staff became necessary. And indeed, almost every major web company failed to make any real money until they built up a massive sales staff. We’re hearing this all the time now, for companies like Groupon.
Taking a tour of the Groupon offices, Leslie Stahl commented on 60 Minutes this week, “Where most websites rely on algorithms, Groupon relies on actual human beings.” Ignoring the fact that Groupon, like Google, is so big that additional advertising revenue growth is perhaps limited by the actual size of the economy, the only way Groupon is going to grow is either by increasing the revenue per salesperson, or increasing the number of salespeople. Groupon has 10,000 employees, as mentioned in the segment. Their revenue, after the adjustments required by the SEC, was reported as $688 million for the first six months of the year. By my calculations, that works out to a revenue of about $11,500 per employee per month.
Google fares better, and the ad-serving algorithm at the core of AdWords gives them a noticeable boost in revenue per employee. Yet it’s a massive company, and subsequent products have not proven as algorithmically friendly. The best guess for the number of employees at Google is their self-reported “more than 20,000.” We’ll be generous and call that 20,000, as the fewer employees the higher the revenue per employee per month. Google’s Q3 2011 revenue was $9.7 billion, or $3.233 billion per month. That works out to $116,000 per employee per month of revenue. Facebook fares a little better, but not by much: $4.27 billion in 2010 revenue, 3,000 employees, or $118,600 per employee per month. While I can’t give official numbers, I can tell you anecdotally that at my humble agency, I would have packed up and shut down if our revenue per employee was so anemic.
Let’s compare that to McKinsey. McKinsey had somewhere around $6.6 billion revenue in 2009 and 17,000 employees (it’s a private company, so they don’t have to report their revenue). This works out to $37,352 per employee per month. Seems a great deal better than Groupon. Heck, let’s take something public.
Let’s look at the advertising services conglomerate Omnicom. Omnicom had 2010 revenue just over $12.5 billion, with 65,500 employees. This works out to $15,900 revenue per employee per month. Omnicom’s market cap is $12.82 billion, or just over 1x revenue. Coincidentally, this is nearly identical to Groupon’s market cap. Except it has something like 20 times the revenue.
Indeed, looking at the Fortune 500, in terms of most profitable companies per employee, Google, with ten times the revenue per employee of Groupon, doesn’t even make the list. Let’s take GE, with $150 billion+ in revenue and 287,000 employees. This gives us a monthly revenue per employee of $43,500, and a market cap of around 1.25x revenue. Facebook and Google may boast a revenue per employee some 3 times that of GE, but how does that justify a revenue multiple of 21x instead of 1x?
The argument, of course, is that these tech companies are positioned to scale in the future, and that their market cap represents that. They are making new markets, so it is easier to grow than, say, Omnicom or GE. So. Let me ask you this: Which company—Omnicom or Groupon—has a better ability to scale? GE or Google? How is Groupon going to scale in some sort of algorithmic way that justifies its current market cap of 18x revenue? Which company is going to hit a wall in terms how big their market is? Where is Facebook going to find that billionth user? Who is going to have a harder time hiring more and more and more people? Can Groupon find developers and salespeople more easily or cheaply than Omnicom can find account people and art directors? Who has more relentless competitors? I see no plausible scenario where Groupon or Google can grow any faster, further or profitably than GE or Omnicom. All the companies are already global, and any regional expansion Groupon may have yet to exploit still means feet on the ground, more employees, and probably chasing less revenue in less wealthy markets.
The myth of algorithm made a lot of sense. It would be lovely to see it work. Indeed, the Fortune 500’s most profitable company is financial services company INTL FCStone, which provides trading services for clients primarily in the commodities business, where presumably some algorithms are at play. They boast a stunning $5.3 million revenue per month per employee. Their revenue is $423 million. Their market cap $465 million, or just over 1x revenue.
Algorithms can be powerful. But with our tech titans, the idea that more money comes from fewer people appears to be nothing but a myth.