The Real Cost of Freemium

Chris Anderson's book Freemium has caught a lot of attention in the high tech and media industries. When your marginal cost of goods or services is virtually zip, the idea that you could have all the customers you want by giving away product or services to a large group and converting a small portion to paid premium offerings sounds like the marketing silver bullet. Only, as I've been hearing from a number of companies in this arena, it isn't. Not that they'd say following the business model has been a mistake. But the mathematics of converting enough customers from free to fee is far from certain and a mistake can be precarious.

"You can't ignore the cost of the free use," says Aaron Levie, CEO of collaboration services vendor Between development, hosting, Internet traffic charges, and general business infrastructure, "free" services are anything but. For, that means 2.5 million free users and somewhere around 50,000 paid users, or 2 percent. When I asked him what level of conversion he'd need for profitability, he answered, "We have the good fortunate to working with investors in the area. It lets us invest more aggressively than if we were purely bootstrapped. It afforded us to not have to look at the bottom line too closely." In other words, more.

He says that he knows anecdotally of examples where freemium models work well. But it's still something where "everyone is still kind of learning as they go."

The free use actually becomes "subsidized marketing for the paid service," according to Ranjith Kumaran, CTO and co-founder of YouSendIt. In fact, he thinks that freemium may become necessary on the web simply because "without traction, it doesn't matter how good your product is or how much you spend to market it." A company needs critical mass if it want to start growing significantly, he thinks. Right now YouSendIt has 10 million registered users and 100,000 paid subscribers, not including corporate accounts or pay-per-use. Overall, the conversion rate from free to paid is over 3 percent. It will take less than 5 percent paid customers to get the company to profitability. "Today, if I had 50 percent more subscribers, we'd be darned close to break-even."

Those low conversion numbers may sound heartening to many small software companies, but they shouldn't be. As Levie notes, the financial burden is not the free users, but the cost necessary for either free or paid users. A second way for a company to look at the model is not as a conversion rate, but as the raw number of paid uses necessary to support the underlying infrastructure. Instead of thinking, "We only need X percent of people to convert," think of it as having to sell 50,000 or 100,000 people on a product or even on a continuing service.

Where the conversion rate becomes useful is to get a realistic sense of how many people would have enough need to pay for the more extended version of a product or service. For example, I spoke with Milo Sindell, CEO of Knowledge Genie, which my colleague David Weir at BNET Media wrote about. The company is self-funded and needs about a thousand paying clients to make it work. That's a low absolute number, but he's guessing that a quarter of free customers will wind up paying $24 a month. When I told him what typical conversion rates are for companies using the freemium model, he was taken aback.

It might be that Sindell is in the right market, aiming at people who teach or coach in given areas and who would like the equivalent of a book to sell, but who think they can do without traditional publishing. However, even though looking at raw numbers is important, the conversion rate doesn't become unimportant. It's a measure of how long it will take to get to profit at the current rate of attracting users. And getting all those people so the few percent can find it useful enough to pay can take a lot of time.

Finally, there may be areas where the economics of freemium simply cannot work. To be successful, a company must find a niche in which the pressures on at least a small portion of the total users are great enough that they become willing to pay for extended capabilities that offer value they come to realize they need. That alone may eliminate most b-to-c plays.