For a while, freemium business models--meaning, giving away a portion of your online services while charging a small percentage of customers for premium versions--were all the rage. Such companies as LinkedIn, Dropbox, and Skype seemed to do well. Wired editor Chris Anderson published the definitive book about it in 2009. (Yes, it cost money.)

Fast forward to 2012, as The Wall Street Journal notes that some companies get trapped with "higher operating costs and thousands of freeloaders." A company called Chargify, offered bill management software. If you had fewer than 50 customers a month, you didn't have to pay. More than that, and the cost was $49 a month. But the company was speeding toward bankruptcy on the freemium highway. It took a big change--requiring all customers to pay at least $65 a month--to get out of trouble and finally become profitable.

Freemium has become popular because cloud computing services have made it possible for start-ups to avoid major fixed costs and focus on incremental costs that have the advantage of the service provider's scale. It has made it relatively cheap to get into business, and even cheaper to attract more customers by using the magic word "free," saving significant costs in marketing. Just get enough people who need your paid offerings and you can do nicely.

But that's where things get sticky. Entrepreneurs tend to be an optimistic lot. Freemium models, on the other hand, are unforgiving. Over the last few years, I've spoken with a number of companies that have successfully used freemium strategies. The reality is that conversion rates from unpaid user to paid customer are typically very small: low- to mid-single digits.

The difference between a rate that will just work and one that will be a sure path to financial disaster is tiny, even when you're being realistic. I remember talking to a new start-up that was planning to use a freemium model. The owner told me that he had planned on something like a 25% conversion rate.

It's impossible to structure realistic projections if you are so wildly off what is likely to happen. This is similar to the many people who write business plans with the assumption that they can take 1% of their market. In most cases, that's highly unlikely, but people do it by rote without considering whether the assumption is valid.

In addition to being realistic about the potential conversion rate, a clear-eyed look at costs is just as necessary. Entrepreneurs who have tried cloud computing are warning that it is all too easy to use too many resources, forgetting what you've already ordered, and suddenly find costs spiraling out of control.

In other words, freemium companies using the cloud for operations are particularly vulnerable at both the top and bottom lines. Freemium might be just what your business needs to jump start and thrive. But before you jump into the water, make the proper measurements to be sure you don't unexpectedly hit a surprisingly shallow end.