A technical journalist has decided to argue with me about enterprise-software market research and business models and my warnings not to depend on the superficial writings of technical journalists in your information technology (IT) investing.
The nut of the argument is this
- InformationWeek reports that it is a new phenomenon that so-called enterprise software suppliers such as IBM, Oracle (ORCL), and SAP are really in the services business. It has written about a half dozen articles along these lines in 2010.
- The journalist thinks software companies becoming services companies is only a 10-year-old phenomenon and feels therefore -- I guess -- that I should not call out InformationWeek for saying it is new news. (I actually don't understand the substance of the journalist's argument because he and I are closer in terms of analyzing the time periods involved than he is to InformationWeek. Going back to his post, I guess he's arguing with me because he doesn't know me. I don't know him either but that is an odd thing on which to base an opinion.)
- I not only think, as the journalist does, that the trend is old news but I also think that in fact the business the enterprise-software suppliers wanted to be in right from the beginning was services. It's just an historical blip that the industry licenses software the way it does, going back to unnecessary manipulation of the market by the U.S. government in the late 1960s (speaking of which, congratulations to Steve Mills). It's just that if you are starting from scratch, it takes a while to get there. Oracle figured that out 5-8 years ago and starting acquiring services revenue streams like crazy. The former enterprise software company formerly known as CA after being known as Computer Associates figured that out in the 1990s.
My original blog post (and InformationWeek) was talking about the entire license to services revenue ratio for the typical 1980s/1990s-era independent software vendor startup. That historic ratio is about $2 in services to $1 in license. Do the math assuming client one buys a license in year one, pays for training and implementation services that same year, and then keeps it "maintained" in subsequent years. The first year that's 80 cents in license revenue, 20 cents in maintenance, and whatever the client spends on training/implementation. Then assume a second new customer does the same thing the next year and a third the third year, etc. Then it's 80 cents in license revenue, 40 cents in maintenance, etc. in the second year, 80 cents in license, 60 cents plus the third year... you get the idea... until the cross-over point. This assumes satisfied customers; about 90%-plus are. (Note: as discussed in other posts, "maintained" is in quotes because maintenance is about content, not fixing things.)
Perhaps the journalist was just looking at the historic maintenance/license split -- not counting professional services such as training and implementation -- but even that should approach a 1:1 ratio very early in a successful traditional enterprise software supplier's business model. For SAP, the cross over came in 2001 according to its SEC filings (see 2002 20-F), less than 10 years after its big push into the U.S. began. Revenue recognition rules changed around that time as well; apples to apples, the cross over might have been years earlier for SAP. As late as 1999, SAP was still doing nearly $200 million a year in R-2 maintenance.
For Oracle the crossover was 1997 (and as I said in the above linked April 2010 post you must take into account that Oracle changed its business model to this approach in the early 1990s). For JDE, the crossover may have happened prior to its 1997 IPO (if not, the company was already approaching the 1:1 ratio at that time according to its S-1). Another thing that might confuse the journalist is that maintenance service was often termed as being "licensed" in the 1990s (see Mapics 10-K of 1997).
As I said in my July and April posts, Microsoft and others -- including some of the components of what is now Infor -- did not use this model but depended on partners to perform services. But even they eventually acquired the partners and adopted the services-centric model.
I am sorry journalists all but this is just how it was done "back in the day." To pretend this is new news is just not accurate. It makes it sound to investors as if the leading enterprise-software suppliers have taken some kind of recent defensive posture when what they are actually doing is executing well against a long-time business model. I know it's a boring and therefore hard to call news -- which is what journalists want to write about -- but that's the way it is.
Of course with a Software as a Service (SaaS) revenue flow, there is no distinction between license and maintenance (and hardware and admin costs, etc.--see above vis a vis Steve Mills). I called that business model "back to the future" (meaning the 1960s and 1970s) in my 2001 research of the same name on this subject.
Today, technical journalists inaccurately call it cloud computing.
-- Dennis Byron
(no financial interest in the companies mentioned except for the $12 a year I pay Microsoft for Office)