Where to for revenue recognition?

Fellow Irregulars Phil Wainewright and Jason Corsello are batting around the implementation services revenue recognition issue following Taleo's restatement of earnings. Phil uses it as a reason to take a sideways swipe at the on-premise vendors:

I wonder what the impact on SAP, Oracle and Accenture’s financials would be if they couldn’t recognize all their implementation and license fees until the application had been up-and-running for a three-year stretch? Ouch!

while Jason believes this might allow Taleo to benefit:

I would expect new implementation partners to come calling to Taleo.  This could be great news that allows them to build deeper channel relationships with multiple tiers of services partners.  Not that Taleo is doing a horrible job today (in fact they probably do implementation better than most) but services could potentially become a differentiating factors if they could balance the demand on rapid implementation with long-term client success complimented by more specialized services.

At one time, the software industry had a certain notoriety from playing revenue recognition shell games. One only has to think back to the days when CA regularly announced rule changes much in the style of Rollerball. We now know where that ended with former CEO Sanjay Kumar drawing a 12 year stretch at Club Fed in 2006. Much earlier, the then PeopleSoft got itself into an SEC tangle over R&D expense shifting of some $500 million through a special purpose vehicle. If the Taleo story is another example of the industry cleaning up its act then all to the good. I have reservations.

While Phil is right to point out that this apparent rule change further complicates the saas revenue model, the decision to rateably recognize implementation revenue over the life of the initial contract is wrong headed. In my view, Taleo should have stuck to their guns. It is illogical because it fails the fundamental cost and revenue matching principle inherent in generally accepted accounting principles.

Using Phil's logic for instance, you could argue that services provided by on premise vendors are to be recognized over infinity because application licenses are usually couched in terms of a perpetual right to use. No-one is going to agree to that any more than they'd agree to the cost of filling your car with gas as something to be amortized over the time it takes between refills. That's how silly the logic can go.

The wider point about disparate treatment is worthy of discussion. When vendors treat revenue in different ways then it makes like for like comparisons difficult. This matters to the financial analyst community but should also matter to buyers who should be evaluating on vendor viability. Complexity breeds suspicion and I can imagine that customers might be baffled by these artificial machinations. More to the point, if revenue is to be recognized rateably across the board then I would expect deal makers to extend the logic to include terms of settlement. In doing so, it would expose the logical fallacy to which Taleo has been forced to submit.

Can you imagine a vendor willing to accept settlement for implementation services on the same terms as its online service contract period? It's not going to happen because the disruption to cash flow would be too great. And what happens when a company decides to switch via a get out clause? Using the same logic, the vendor would be entitled to view the contract as terminated and accelerate deferred revenue.

Even so, both Phil and Jason's arguments intensify the light on saas differentiation. Saas vendors only get paid for what they deliver. In those terms, failure is less likely. That's all to the good.

PS - unlike Phil, I used to be a practising accountant. What do I know?