The software industry seems to take a cavalier attitude to its user community when it comes to software upgrades, determining the frequency of replacements solely on supplier wants, rather than user needs.
With Microsoft's new volume licensing policy getting its customers up in arms around the world - from New Zealand to Peru, let alone the UK and USA - the question IT users ask is, "Does it have to be like this?"
There is certainly a long tradition to overcome, says Roger Marshall, information systems director at the Corporation of London. "IT people are often accused of being too much in the pocket of the IT industry, rather than acting primarily in the business interests of their employers - they have always run that risk."
"Generally speaking, users are too dominated by what suppliers want," he says.
What suppliers want, of course, is forward revenue growth. Setting aside revenue streams from services, this comes in two ways - selling existing products to new customers and/or selling new products to existing customers. When the market for the former is saturated, the only way to satisfy shareholders baying for growth in revenues is the latter.
In other industries, where a purchase is a purchase and that's that, suppliers have to come up with real improvements in the next generation of products that customers can see and value or they won't buy them. But because software is licenced on a "restricted right to use" basis, not sold outright, suppliers have far more control over the actions of their customers because they have far more leeway in altering the conditions of licensing, irrespective of any actual improvements to the product itself.
Whereas in boom times users may not have bothered too much about this situation, "We were all lulled into a false sense of security," says Marshall - now that times are harder users are starting to ask, "What are our options - other than taking the easy line pushed upon us by the suppliers?"
But times are also harder for suppliers, which therefore are keener than ever to handcuff users to the upgrade treadmill, so tightening the vicious circle.
Three aspects most infuriate users:
- The frequency of upgrades
- The marginal improvement to the product
- The greater cost of the new version compared with the old one.
Why should they be expected to pay more for something they don't need in the first place, they ask?
Even with latest upgrades, say, coming with Internet features, these may still not be enough to produce a positive return on investment when it comes to a business cost-benefit analysis.
Generally speaking, when it comes to upgrades, "New versions either don't improve on business benefits or provide only marginal improvement," says Marshall. Users, he says grimly, are caught in a "dance of death" it is difficult to break out of.
One user who is trying to break out, or at least slow the tempo, is Alan Paul, business information manager at Marshall of Cambridge Aerospace. He is adamant that if anyone is going to be in charge of the software upgrade cycle, it should, wherever possible, be him.
He is forthright about his upgrade strategy: if existing software meets business needs, don't upgrade it. He feels strongly that for all the bells and whistles piled into new releases, very little actually pays off for business. That is true even for existing software, he says.
"Users only use about 5% of existing desktop capabilities," he says. So if even the Windows 98 generation is under-exploited by his 1,000 or so PC users, how much more will they get out of the XP generation? For users who "just want to do word processing, e-mail and spreadsheets," the Windows 98 generation works just fine. So much so that he is still rolling it out.
"We're still installing Windows 98 on PCs, and would like to go on doing so for another three or four years - although we probably won't be able to," he says. "But the Windows 98 generation will be our desktop standard for the next 18 months. We've just put out 250 PCs this year with Windows 98 - we've deliberately not gone for XP."
However, when he does upgrade finally, it will be straight to XP. "We've bypassed Windows 2000," he says. "Eventually we'll start rolling out new PCs with XP - but only once we've sorted out what applications run on it. It will mean new hardware, because our old PCs won't run XP."
Until that changeover is complete he will, he knows, have two platforms running. "We'll have Windows 98 and XP so we'll be supporting two different operating systems," he says. "Yes, that's an extra cost, but it's still less than buying 600 new PCs [to run XP] now."
Paul is in no hurry to go for XP. There is not much visible business benefit, and even a clear disadvantage from his point of view. "We have a couple of applications [we still want to use] that don't run on XP, and we also don't want users having the kind of remote-control that XP allows - we like to control things centrally," he says.
The trigger to move to XP will be application driven, he says - and have nothing to do with XP itself. "If we need a new application that only runs with XP, at that point we will go down that route."
That avowal points to yet another upgrade bug-bear for the user community - the three-legged race effect. Because application software is dependent on system software, at any time the dominant system software supplier can "whip in" the application software suppliers to keep up with the latest system software.
Although it is Microsoft that is the current focus of user ire and obloquy, thanks to its licensing proposals the upgrade treadmill extends far beyond Redmond. "There's a real feeling of being taken to the cleaners and paying for nothing - or almost nothing," warns Marshall.
"We've been using PeopleSoft 7.5 for our order management, finance and project finance system," says Paul. Although it works fine and he is very happy with it, "Version 7.5 won't be supported after next year," he says. "We'll need Version 8."
But Paul does not want Version 8. It may be Web-enabled, but that's not enough to sway him. "You can make out the business case for moving to Version 8 on the back of a fag packet," he says dismissively. "There's none."
What there is, though, is a bill for £500,000. Ironically, this is not heading PeopleSoft's way. "We'll get Version 8 'free' because we pay an annual support licence," says Paul. The cash will be for the new hardware that Version 8 needs to run on.
From Paul's point of view that is £500,000 to stay where he is, in terms of the benefits the company is getting from PeopleSoft. He intends to continue getting those benefits - but without the new version, and without an expensive hardware upgrade.
"We'll run Version 7.5 unsupported," he says. "We've got enough experience." What will tip him over to Version 8 will probably be the XP upgrade when it comes. "The whole thing is very complicated because there are so many dependencies," he points out.
One key element to Paul's decision to continue with pre-XP desktops is that he does not buy site licences. "We buy all PCs with Microsoft pre-installed by the OEM. We never went for a site licence," he says.
Though that means he has to pay for upgrades, the crucial advantage is that it leaves him in control of the replacement cycle. He can replace PCs, and the software running on them, progressively, as and when there is a business need for them. "We've bought PCs as we've gone along," Paul says.
As well as greater control, he says this also spreads the costs more digestibly - buying 20 or so a month is more easier than a big-bang replacement. "I'd say, per seat, for about 1,000 PCs, it probably works out cheaper than a site licence," he says. "If I wanted to replace my software every two years, then it would be cheaper to have a site licence," he allows.
But Paul's point is that he does not want to replace software every two years. "A desktop should last five years," he asserts. "My own PC is three-and-a-half years old and it's perfectly OK, and the hardware is capable of lasting a lot longer than three years." He only considers new PCs for users who have to work all day at a machine, and would therefore see a benefit from something like faster-loading software.
When PCs are replaced, the old ones are downgraded to "intranet PCs" rather than personal ones. They become available for general use, such as when a visiting engineer needs to access the maintenance manuals held on the intranet. Paul has several hundred such machines.
Paul is ruthless when it comes to assessing the business benefit of new technology. If an upgrade fails to contribute to the corporate bottom line, and can be avoided or deferred, then it is. This does not, he emphasises, make the company a technology dinosaur. He will - and does - happily embrace the latest technology if it brings clear and calculable business benefit.
Telecommunications is a prime example. By spending about £8,000 on hardware, the company now has the use of a virtual private network (VPN) that will communicate to all its remote sites around the world, and complies with Ministry of Defence security standards.
The business benefit is incontestable. "The VPN is saving us £100,000 a year," he says. "Our communications cost for overseas sites is £50 a month." That compares, he says, with a cost of £50,000 for a six-month frame relay link to Atlanta the company once set up.
"We're not frightened of the latest technology," says Paul. He is, however, prudently wary of it - and not just because there is no clear or calculable business benefit to having it. It can be technically risky as well. Paul has just emerged from a year of being, effectively, a beta site for clustering technology which was not sufficiently robust for general release, he says, despite the enthusiastic claims of the salesman.
"It was sold to us as the latest technology, totally resilient," he says. "We believed the salesman - but for the first year it was totally unreliable - it was just not ready for clustering. It's working fine now - the service packs started arriving late last summer - but it was not up to it for a year."
"We were an unwitting and unasked beta site."
How to avoid being forced to adopt a pace that does not suit you
- Never upgrade unless you can see a business benefit or an easier or better way of doing existing work
- Only upgrade system software when you need to upgrade a business-critical application which meets the criteria above
- If there is no business benefit, opt for the advantages of stability
- Be prepared to support two generations of operating system simultaneously if this is necessary to avoid big-bang upgrades
- Be prepared to run unsupported application software if the supplier no longer supports older versions that do not cost-justify replacement but ensure you have sufficient in-house skills, and understand and accept the risks of running unsupported software
- If you install new versions, users will assume it improves their jobs. If - or when - it does not they will ridicule you for installing it, and resent having to waste time learning it
- Understand where your total costs will fall - a software upgrade "free with maintenance" may require an extremely costly hardware upgrade to run on
- Consider buying individual desktops from OEMs with pre-installed system software; provided that you do not intend to upgrade every two years, this can be cheaper than buying a site licence, and give you far more flexibility and control, as well as spreading replacement costs more digestibly
- Use replaced, but paid for, old PCs as lower-spec, general or single-use workstations for mobile staff
- Be enthusiastic for the latest technology - but only where the cost-benefit justification is unequivocal
- Be wary about the latest technology. Being too early an adopter can make it not worth while at first
- Not all latest features improve end-users' jobs. Web-screens, for example, can perform slower than client-server systems
- When it comes to new technology, keep it simple and do not believe the software salesmen
- Accept that managing the upgrade cycle in your favour, not the suppliers', will require management time and effort
- Do not roll over and let the upgrade juggernaut crush you.