Is Conventional Software Built on Bubble Economics? (Hint: It’s All About Recurring Revenue!)
Posted by Bob Warfield on September 27, 2007
I wonder sometimes whether the conventional software model has ever really flourished except during bubbles of one kind or another. Look at Big ERP. They had the benefit of several major bubble waves: the rush to switch to client server architectures, the Y2K debacle, and then right on the heels of Y2K was the Internet Bubble. Would these businesses have grown nearly so large without any of those bubbles? Maybe SaaS would’ve gotten here sooner. The choice between the two is a Tortoise (SaaS) versus the Hare (Old School ISV) race.
Phil Wainewright was taken aback by it, and in retrospect, it was unfair to drop that out there without giving it a blog post of its own to explain. Before I do that, let me summarize some remarks others have made that are pertinent.
Phil runs with the premise for a bit, likening the purchase of conventional software to subprime mortgages, suggesting companies must then rue the purchase when normal times return, and further suggesting that in a recession, conventional revenues will collapse while SaaS revenues will “power ahead”. While Phil juxtaposes Larry Ellison’s denouncement of SaaS (where’s the money in SaaS?) briefly against my comments, he finishes with a great Talkback post where the poster says the only reason Larry doesn’t pursue SaaS at Oracle is because its too hard to reinvent the company, which is exactly what Steve Singh says (one of the few guys who has successfully reinvented a good sized company). To all of this, I do say that conventional revenues are much harder hit than SaaS revenues, but not all conventional software companies are hit the same in recession. I will explain in a moment, but it all boils down to recurring revenues. Before leaving Mr. Wainewright, I suspect after reading his piece that on balance he sees there is some truth to the idea that conventional ISV’s are built on bubbles.
SaaSWeek picks up on the theme. Their penultimate question and conclusion:
“Is SaaS an inherently stronger model that is destined for eternal greatness by inherent design? Nope.”
Their thining behind this conclusion is that they feel that dealing with the scale issues of very large businesses as well as their flexibility requirements are too hard for SaaS companies.
They put scale as, “Anyone want to run GE’s global payroll using SaaS?”. I think they’re dead wrong on that one. ADP handles companies that large all the time for payroll and they’re one of the biggest SaaS companies out there. I’m familiar with the scale requirements of these kinds of organizations–Callidus Software pays sales compensation for some of the largest sales forces in the world for companies like Allstate, Sprint-Nextel, and many others. SaaS has no problem there.
Flexibility is a much harder issue, and one that I will have to wait for future postings to address more fully (flexibility is also related to the SaaS+Open Source discussion I need to address separately too). For now, let me just say that it is an area where many SaaS vendors have dodged the bullet by simply refusing to offer much flexibility. That works for applications where flexibility doesn’t add much business value, but even Salesforce is investing heavily in making their solution customizable. However, there is no reason why SaaS vendors can’t strive to deliver flexibility, it’s just harder to do because you need sophisticated meta-programming capabilities on top of your multitenant architecture. Trust me, we’ll drill down on that later.
SaaSWeek goes on to say:
A manufacturer of canned corn can be fairly certain their business won’t fluctuate much in terms of workforce, product innovation, economic cycles and so forth. Why would they spring for a SaaS solution when they can call a local ERP vendor, eat the initial cost but then run on the same system for 15-20 years with bare-minimum, in-house maintenance?
Here I think we have an apples to oranges comparison that SaaS vendors deal with frequently. You can’t compare the cost of maintenance of On-Premises software against the (apparently) higher cost of SaaS. The SaaS cost much more closely reflects Total Cost of Ownership. Maintenance only entitles you to upgrades, which our canned corn friends have no interest in, and the ability to call Tech Support with questions. TCO for the canned corn gang is going to include keeping the data center running, doing backups, replacing hardware that fails during that 15-20 years (any of you run a 15-20 year old computer), installing patches when problems come up, paying the salaries of the IT people that do that stuff, and so on. When you look at the IT salaries in particular, SaaS comes out way ahead in nearly any sober calculation. Look at Steve Singh’s numbers he gave in the interview for how much he saves folks with his software. Touching back on scale for a moment, Steve Singh’s Concur sells to very very large organizations just as well as small ones and its all SaaS. I would argue the closer your organization is to needing something that has the best possible economics and doesn’t need uber flexibility, the closer you are to wanting to insist on SaaS.
BTW, in a recession, aren’t those the qualities you look for: best economics and skip the bells and whistles? Don’t you want to live with what you can get by on until there are better times? Are you really going to invest in a boil-the-ocean IT rework of major infrastructure when everyone is hunkered down in the nuclear winter? And doesn’t SaaS let you get by paying for just what you need as you need it, and leave you with the option to easily change your mind later when times are good and you do decide to boil the ocean? Do you change paradigms like ripping out your mainframe stuff to install shiny new client/server ERP during a recession? Those are some gut arguments in favor of SaaS during lean times.
Now let’s get back to recurring revenue. The lower TCO argument tells why customers want to buy SaaS in bad times instead of conventional software. Recurring revenue is what you need as an ISV in bad times. Every software company faces a potential downturn in new customer bookings during bad times. That includes SaaS companies. But here is the difference. With perptetual software, every quarter is a new ball game. What you sold last quarter is recognized revenue and no longer matters in the current quarter. If the bottom falls out and you sell very few new customers, the perpetual software company is busted. The SaaS company looks a little different. The reduced new customer sales are riding on top of recurring revenue. So the percentage change looks less. Let’s see it graphically:
This compares a conventional new sales model versus a SaaS recurring revenue model through good times and bad. We assume that SaaS recognizes 1/4 of each new sale each period, but that it gets to keep recognizing 1/4 a period forever (e.g. no churn). We assume conventional recognizes the entire new sale in the same period. We start out selling $2M a period (perhaps a quarter), and we three quarters of 30% growth followed by a period of decline, and then a period where growth comes back. The SaaS company has the same problem–it can’t sell as many new customers during bad times either, but it falls back on the cushion of recurring revenue.
Now I ask you, which company would you rather have invested in? Which one would you rather be working for? Do you see why I wonder about whether the success of big conventional software isn’t built during bubbles when abnormal growth rates fuel a meteoric rise? Clever readers will surmise the next shoe I want to drop. It’s an old shoe too: big conventional ISV’s have a huge advantage over small conventional ISV’s. Yes, it’s all the old stuff about brand and yada, yada. But, to make it worse, they have a lot more recurring revenue! In fairness, all conventional ISVs have some recurring revenue so the graph doesn’t look as bleak as I’ve drawn it.
Conventional ISV’s can gain recurring revenue from at least 2 sources: maintenance revenues and professional services. Even if you have a downturn in new license sales, you still have some recurring revenue cushion. But the biggest conventional ISV’s like Oracle and SAP have been building up their maintenance base for many years until it is substantial. And yes, Virginia, the bubbles I mentioned around client/server adoption, Y2K, and the Internet Bubble, all tremendously fueled that build up. For those not familiar, it is customary on Big Enterprise Software to charge an annual 10-20% maintenance fee. All big responsible companies pay their maintenance if they’re using the software (and some pay even if they aren’t using it, but that’s another story)
Professional Services is another source, albeit a less efficient one (because it is of shorter duration and lower profitability), of recurring revenue. If you’re engaged in a long term installation or if you’re called back into the installed base to install an upgrade or do further customization, you get to charge for services over a period of time. It’s likely not as long as a SaaS or maintenance contract, but it is significant. That’s why it should be no surprise that when times are tough selling new licenses, companies hunker down around their services organizations.
Which brings us back to SaaS, where Software is the Service. It’s all Services. It’s all recurring all the time. This is a beautiful thing!
So there you have it. The economics of recurring revenue streams and why they mean SaaS companies have yet another long-term structural advantage over conventional software. They can continue to show growth even when things are quite difficult, so long as they can sell more than their churn (i.e. customers who leave and don’t come back) in new recurring revenue contracts. In a future post we will consider churn at SaaS companies and some of the factors that lead to churn as well as strategies for minimizing churn.