If you want to understand the strange dynamics that drive technology procurement in large corporations then the question of whether such organizations should use SAS or R will reveal most of them. SAS is often described as a legacy analytics platform by the new kids on the block who want a share of the action. What this really means is that SAS has been around for several decades and is deeply embedded into the culture, technology architectures and mindset of many large organizations. The fact it is very expensive really isn’t that much of an issue to a multi-billion turnover business – a few hundred thousand dollars or maybe more is little more than loose change to some of these corporations. Often their operations depend on the hundreds of SAS routines they have developed and run on a day-to-day basis. Changing would be unthinkable. It’s a sort of arthritic condition, as applied to businesses. Too old and set in their ways to change – and they most certainly won’t.
When it comes to a business that has no SAS legacy the decision should be fairly straightforward. R is free and SAS is expensive. But this is where FUD (fear, uncertainty and doubt) comes into the equation. Most of the people who make decisions that involve decent amounts of technology spend, typically have very little idea of what they are buying. It’s called Putt’s law, and it says that those who understand have no budget, and those who don’t understand do have budget.
So the budget holders feel insecure and are fair prey for the big suppliers that can offer support, consulting and maybe pre-built solutions – for fairly large amounts of money. R on the other hand looks untamed and doesn’t come with any slick marketing material. Companies such as Revolution Analytics are trying to change this perception, but it’s going to be an uphill struggle.
At a technology level the usual insecurity associated with R is that the large collection of packages are written by individuals (some of whom lead in their field) and as such there is no warranty that comes along with them. But buyers should look at the small print from suppliers such as SAS and IBM – they may be surprised to find there is no warranty there either. Most people who know SAS and R tend to prefer R because of the power of the language. A purely rational comparison would say they both do similar things, whereas one is free and the other expensive – so pick the free one. But despite our pretense at rationality, in business the primary driver is fear – fear of being wrong, unsupported, in a minority … It is much safer to have made a bad decision along with a hundred thousand other businesses than make a bad decision and feel isolated – and so ‘me-too’ will generally win.
Selecting R really is no more risky that selecting SAS, SPSS or any other big commercial name. It’s all a matter of perception – which is why software companies spend up to 70% of their revenue on marketing and sales.
Finally there is one fact that may eventually determine the outcome of this deadly embrace. Nearly all suppliers of analytics tools (Oracle, SAP, FICO, Angoss …) have integrated R into their offerings. It’s becoming the lingua franca of the analytics world. You may find that you are using R anyway – like it or not.