Arik Hesseldahl

Recent Posts by Arik Hesseldahl

The Red Flags That Were Obvious — To Some — In the HP-Autonomy Deal

The one big, glaring question that will probably never be fully answered in the still-developing HP-Autonomy scandal is this: If buying Autonomy has so obviously turned out to be a bad deal 15 months after it was first announced (and don’t forget that Autonomy’s founding CEO Mike Lynch rejects everything HP says), why weren’t there any red flags that could have warned HP before the deal was consummated?

It turns out there were, and a few smart short-sellers — the most obvious one is Jim Chanos of Kynikos Associates, who in the last 24 hours has been hailed as a bit of a god of short-selling wisdom on CNBC — who read the situation correctly and made money.

There are others who saw troubles at Autonomy that should have occurred to HP’s due diligence team. One is John Hempton, of Australia’s Bronte Capital. In a blog post this morning he sums it up this way: Good software companies tend to have low receivables and higher unearned income on their profit/loss and balance sheets. At Autonomy, it was the reverse.

Here’s a little Accounting 101: A receivable is a debt a company is owed and expects to be paid. If you buy a car from your neighbor and pay him half now and half next month, he’s counting on the fact that he’s going to be paid that other half next month. This is a receivable, and on a corporate balance sheet it goes in the assets column. This is how it tends to work for durable and tangible goods, which software is not.

Software is usually sold for cash up front, and then any further payments tend to come from either service and support on an ongoing basis or, as in the case of companies like Workday or, as subscription revenue. In either case, the company selling it can’t report the ongoing payments as income because it hasn’t been paid yet, and since there are no tangible goods involved, the risk of not being paid is higher. The customer could go out of business or cancel the subscription or simply stop paying. As such, the expected future payments can’t be counted as assets, and in fact have to be accounted for as a liability, essentially money that might be lost. Sometimes they’re called “deferred revenue” or “unearned income.” Still, these numbers can and often do provide a strong indication of future revenue, which is why cloud software companies like Salesforce, Netsuite and Workday report them regularly.

So here’s how Hempton describes what he says was the most glaring problem that should have been obvious to the team at HP conducting due diligence on Autonomy:

Sales were $870 million.

Receivables were $330 million — which is four and a half months of receivables.

Deferred revenue is $177 million — just over half of receivables.

This is really perverse for a software company. Software companies sell stuff that is barely tangible — they sell it up front and for cash. They have very few receivables.

They do however have an obligation to service that software for a long time after they sell it — so the unearned income is relatively large (usually a multiple of receivables).

Autonomy was booking as income lots of cash it had not received (which is why the receivables were large) and not booking any obligation to provide future services for that income.

This is prima-facie suspect (and you could tell simply by looking at the balance sheet). All it required was basic applied accounting.

Here’s another warning sign. I’ve talked to a few people familiar with traditional ways of valuing software companies, and they tell me that Autonomy’s operating margins, which grew from 15 percent in 2005 and surpassed 50 percent by early 2010, were too high to make sense.

Autonomy, they say, hadn’t grown to sufficient scale — at sales just shy of $1 billion with a relatively small base of customers — to make margins in the upper 40 percent range realistic or likely. Software companies increase their margins by boosting the number of customers they sell to. The more customers you have, the better to amortize your research and development costs, which are essentially the only real costs associated with creating software.

The thing is, once you make a sale, R&D costs don’t recur. The sale to the second, third, fourth and fifth customers are, aside from any marketing and costs incurred to make the sale itself, essentially pure profit. With a larger customer base, each customer accounts for a smaller and smaller sliver of the R&D cost. And once you incur the R&D cost, it’s over with. Thus, margin and revenue move together.

As one executive of a software company who didn’t want to named observed to me today: “Autonomy didn’t have the scale to achieve the margins they claimed. Once you understand that, it’s impossible to see how what HP paid made sense.” Hm.


Latest Video

View all videos »

Search »

There’s a lot of attention and PR around Marissa, but their product lineup just kind of blows.

— Om Malik on Bloomberg TV, talking about Yahoo, the September issue of Vogue Magazine, and our overdependence on Google