2014 was a very unprofitable year for Twitter (TWTR). The company reported a net loss of $578 million on $1.4 billion of revenue. Even the free cash flow, which benefits from adding back Twitter's massive $632 million of stock-based compensation, was negative, a loss of $120 million. The fourth quarter was no different than the full year, with both net income and free cash flow soundly negative.

Despite these sobering numbers, Twitter recorded record quarterly profits, according to CEO Dick Costolo:

We closed out the year with our business advancing at a great pace. Revenue growth accelerated again for the full year, and we had record quarterly profits on an adjusted EBITDA basis.

EBITDA stands for earnings before interest, taxes, depreciation, and amortization. Twitter's adjusted EBITDA backs out another major expense: stock-based compensation. On an adjusted EBITDA basis, Twitter earned $141 million of profit in the fourth quarter, and $301 million of profit during the full year. Adjusted EBITDA nearly quadrupled in 2014.

If you look at the earnings reports of big, profitable technology companies, you're unlikely to find EBITDA mentioned at all. I looked at the latest earnings releases for Microsoft, Apple, Intel, Cisco, Facebook, Qualcomm, Oracle, IBM, and Google. How many times do you think EBITDA was mentioned?

Not even once.

There's a good reason for this: EBITDA is a mostly useless number. EBITDA was a popular metric during the dot-com boom of the late 90s, and I'm seeing it touted in an increasing number of earnings releases today, largely by unprofitable tech companies like Twitter.

A common argument for using EBITDA is that depreciation and amortization are non-cash expenses, and by backing those out, you get something that's supposed to represent a company's real cash flow. Warren Buffett, in his 2000 shareholder letter, pretty much kills this argument:

References to EBITDA make us shudder -- does management think the tooth fairy pays for capital expenditures?

EBITDA accounts for the earnings generated by a company's assets without accounting for the cost of those assets. Depreciation may not be a cash expense, but it is a real expense, and ignoring it produces a number that carries little meaning. Charlie Munger puts it best:

I think that, every time you see the word EBITDA, you should substitute the word "bullshit" earnings.

To see why EBITDA can be so misleading, let's start a business.

The power of EBITDA
I'm going to start a fictional food truck business. I'll buy one truck in the first year, then add an additional truck each following year. Each truck costs $50,000 and has a useful lifetime of five years, generating a $10,000 depreciation expense annually.

It turns out, I'm not very good at selling tacos out of a truck. Each food truck I own consistently generates $100,000 in annual revenue, but it costs $90,000 to operate, including food costs, wages, etc. Add in the depreciation expense, and my operating income per truck is zero.

No matter -- I want to build a food truck empire. I borrow money each year from the bank to open a new food truck and replace any that need to be replaced, paying 6% interest. After 10 years, with 10 food trucks in operation, things are either going very well or very poorly, depending on which numbers you look at.

In year 10 of my food truck business, the company generated $1 million in revenue, a tenfold increase compared to the first year of operation. It was also a record year for profitability on an EBITDA basis. EBITDA came in at $100,000, 10% of revenue, and it has increased every single year. EBITDA also easily covers the $45,000 of annual interest payments.

That may sound great, but net income in year 10 was a loss of $45,000 thanks to the interest on all of those loans. Free cash flow is also negative, since I spent $50,000 expanding my food truck empire and another $50,000 replacing one of my existing trucks. The company is hemorrhaging cash and, unable to even pay the interest on its loans, is on the verge of bankruptcy.

Calculations and chart by author. Free cash flow ignores effects of changes in working capital, but including this would make free cash flow even more negative.

EBITDA has managed to make this unsustainable, money-losing company look like a success story. Not only does EBITDA ignore the cost of the food trucks, which are required to generate any revenue in the first place, it also ignores interest, which is a real cash expense. EBITDA is in no way equivalent to cash flow.

Putting lipstick on a pig
Twitter's adjusted EBITDA not only excludes depreciation and amortization, which totaled $208 million during 2014, but it also excludes $632 million of stock-based compensation. More than 40% of Twitter's expenses are completely ignored in an effort to produce a number that makes the company appear profitable.

A good rule of thumb for all investors to follow: Always question the numbers management wants you to see. Companies make up all sorts of non-GAAP, adjusted figures in an attempt to make things look better than they really are. Sometimes, these figures are perfectly reasonable. Other times, like in the case of Twitter's magical adjusted EBITDA, they're not.

As Warren Buffett said during the 2002 Berkshire Hathaway annual meeting:

People who use EBITDA are either trying to con you, or they're conning themselves.

Remember that the next time you're looking at an earnings report.