And now this: Jungle logic:Amazon.com ( AMZN) may have beaten revenue expectations and turned in a narrower-than-expected loss for the first quarter, but the real story to some investors is gross margins, one sign of how well run and potentially profitable a company really is. The fatter the margins, the better. And on the surface, Amazon's margins weighed in at an impressive 22.3%, up from 13% in the previous quarter. But dig deeper into the numbers and there are reasons to believe that the real gross margin is more like a paltry 1.2%. The actual size of Amazon's real margins depends on two things. First is whether you include revenue from a series of recent marketing deals with upstart companies that aren't making any money. Short-sellers don't think you should, because those companies are not really part of Amazon. By not including those revenues, the gross margin would be more like 18.5%. More important is whether you include fulfillment costs in the gross-margin calculation. Doing so would chop another 17 percentage points off margins, bringing the actual number down to around 1.2%. The reason this is so important is because like many e-commerce companies, Amazon currently expenses fulfillment costs (the cost of filling orders) as they occur and doesn't include them as part of the cost of sales, which determines a company's gross margin. But, as Amazon itself warns, it may have to adjust its accounting and classification of fulfillment costs as a result of an ongoing debate in the accounting industry over whether fulfillment costs should be treated as a cost of doing business, and thereby included as a cost of sales. Particularly disturbing this quarter is that fulfillment costs actually rose a percentage point from last quarter. Bottom line, according to one skeptic who is short Amazon stock: "They're selling books, DVDs, consumer electronics and everything else for barely above cost." In which case, no matter what the company says, profitability is hardly around the corner. Confess as we must do: An item earlier the other day on StarTek ( SRT) had several important numbers that were off. The column said that the growth in Microsoft's ( MSFT) biz with StarTek had slowed to 16% in the fourth quarter from roughly 100% in each of the prior three quarters. In reality (thanks to poor numbers crunching by a normally impeccable source, I say with a red face) year-over-year sales growth from Microsoft fell to 6% in the fourth quarter, after rising 76.6% in the third quarter, 132% in the second quarter and 105% in last year's first quarter. (Sure, that's worse than the column said it was, but keep reading.) The column also said that StarTek's non-Microsoft business in the fourth quarter had actually fallen by 6% -- the second-straight quarterly decline. Oops, it actually rose by 26%, and there had been no decline in the prior quarter. This is where things get murky, because the column reported that a decline in non-Microsoft sales was "especially troubling" in light of the way Wall Street had tagged StarTek as an e-commerce play. It's e-commerce in part, because its employees are hired to fill e-commerce orders and it is a minority partner with Reader's Digest ( RDA) in a small Internet company. Still, even with the correct numbers, the story doesn't change for several reasons: First, there's the Microsoft connection: Sales growth from Microsoft is continuing to slow. Just how much is hard to say. The company, which reported earnings early Tuesday, doesn't hold conference calls. StarTek Chairman Emmet Stephenson told my assistant, Mark Martinez, on Wednesday that he didn't know the Microsoft numbers because the calculations hadn't been finalized. (Then how could they calculate the earnings? But I digress.) What he did say is that Microsoft's percentage of overall sales is going to be down, "and not insignificantly." Next is the nature of its most-recent quarterly results: Revs were a few million bucks lower than expectations, but earnings were higher. How'd that happen? Stephenson attributes it to higher-margin, non-Microsoft biz, which sounds great, but most of that high-margin biz, according to what Stephenson told Martinez, was coming from doing installations and maintenance for a large telecom company, which he declined to identify. In other words, here's a company that trades at 36 times this year's expected earnings. By contrast, sales for AHL Services ( AHLS), which also outsources for mundane jobs -- including e-fulfillment -- are almost triple StarTek's, yet it trades at just seven times this year's expected earnings. Sure, AHL recently blew a quarter, but to short-sellers, outsourcing is outsourcing, and no matter where their employees go, there's no way to justify a high-tech multiple for a low-tech biz. Transaction complete: Tuesday's 47% decline in Transaction Systems Architects ( TSAI), after reporting lousy earnings, was no surprise to readers of this column, but it was to a team of Credit Suisse First Boston analysts that upgraded the stock a month ago. Tuesday, they conceded that the upgrade was "obviously premature and wrong given performance." But then they kept their "strong buy" in place while slashing estimates for this year and next. Why not drop the recommendation along with the upgrade? What, and risk the prospects of future investment banking biz from a company that has signaled that there are likely to be some deals in its future? Get real!