It’s been said you can prove almost anything with statistics.
Forrester Research has numbers that suggest VARs selling servers to enterprise-sized organizations can get a good idea of their success rate by merely peering in the data centre door and looking at the brands.
Can it be that easy?
Here’s the rationale by report author Frank Gillett: Just over a year ago, Forrester surveyed IT managers at 525 companies in Canada and the U.S. with 1,000 employees or more and asked who they bought from in the previous 12 months.
Sixty-six per cent said they purchased a server from Hewlett-Packard, 65 per cent bought from Dell, 58 per cent from IBM – a fairly even divide – and 35 per cent from Sun Microsystems.
Only 5 per cent of respondents had bought from the next highest vendor, so Gillett stuck with those four for further analysis.
Seventy-six per cent of respondents had bought servers from one of the four vendors. Most (38 per cent) had three of them.
But was there a pattern to their buying?
Yes, suggest the stats: 22 per cent of the shops bought HP, Dell and IBM servers. Eleven per cent of the HP shops bought only Dell. Ten per cent bought all four.
So what, you say. Well, look at the figures another way and they show 61 per cent of HP customers put Dell servers beside their machines, 58 per cent selected IBM hardware but only 40 per cent picked Sun.
Dell buyers were more likely to buy HP (63 per cent) than IBM (56 per cent)
IBM customers were more likely to pick HP (66 per cent) and Dell (62 per cent) servers in their mix than Sun.
More food for thought: 11 per cent of respondents were HP-only shops. Ten per cent were Big Blue-only buyers.
On the other hand no Sun shops were exclusive, and when they bought they preferred HP servers (75 per cent) to Dell and IBM (65 and 63 per cent respectively).
Why the preferences? The manufacturers have different lineups. For one thing, Sun has only been selling x86-based servers for about two years. HP has a broad offering, and aggressive pricing might explain why it’s everybody’s second choice. IBM’s line is also broad, but includes mainframes. Gillett theorizes that if mainframes had been eliminated from the question the figures might change.
He also thinks there are other interpretations.
“Look in the shop to see what they already have and that tells you something about your potential success rate,” he said in an intervew.
“From a VAR point of view, what I’ve just learned is that Sun customers in this data set always buy from another vendor and are therefore more likely to be interested in talking to me than if I go into anybody else’s customer base.”
One IBM reseller I spoke to doesn’t think much of this slicing and dicing. It would be more interesting, he said, to know buying habits of private versus public sector organizations. Government usually buys on price, while businesses take more factors into account.
Gillett, however, said there’s one very interesting conclusion. “What I found striking,” he said, “is the number of server vendors that customers were dealing with. What that tells me is buyers are not trying to standardize on one or two server vendors. They’re continuing to buy based on what works for the application and what the system integrator provides.”
That, perhaps is the real message to enterprise VARs: Listen to the customer.