opinion DOWN TO THE WIRE

Published: 11/4/2002, Volume II2, No. 5800 Page 24

Poor old Alistair Darling, secretary of state for work and pensions, was recently forced to announce an embarrassing delay to the Child Support Agency's new multimillion pound EDS computer system, just weeks before it was supposed to go live.

Because the computer system is not ready, major government reforms to the whole system of child support cannot yet be implemented.

There must have been enormous pressure to meet the planned date, and the decision to delay it could be seen as quite brave. Similar decisions face managers whenever a computer system nears completion.When should you say go and when should you stop? I have managed the implementation of more than 100 new computer systems.

There is frequently user pressure for delay. Sometimes we have moved dates, but usually not.

The outcome tends to be the same. For a large system - and some of ours have literally thousands of users - the system goes in, causes varying degrees of disruption for a few weeks while it beds down technically and users get used to it.A month or two later the system becomes a stable part of everyday working life and problems fade into the past.

The first rule to remember is that it is rarely possible to get absolutely everything done before a system goes live. Possible show-stoppers must be identified and dealt with in advance, but there will always be a list of snags and hassles. No one would want to move into a new house if the plumbing wasn't connected, but a few scratches to the paint work wouldn't put off any but the most fastidious.

If you try to install a major system without any snags at all, you will end up moving the golive time and time again.Moving dates is disruptive to training and re-training schedules, and to planned holidays, and delays benefits. And when you eventually blow the whistle for the start, you are unlikely to be any better off than you were in the first place.

By then the goal posts will have moved and you will be looking at a whole new set of problems.

Of course, if you go for broke without tackling the show-stoppers, you may end up like the Inland Revenue or the Passport Office.

I would bet a fair sum that the CSA had to be thoroughly convinced that the system wasn't going to work, before its boss went redfaced to the minister and asked him to make his statement. So how do you make that decision?

Before going live, think about the worst that could happen and look at your contingency plans.

Could you run manually for a few days? How many days? How would you cope with the backlog?

Would the old system still be available? Would people know how to work if the new system failed? Is the key data safe? Losing all your outpatient diaries would be a major problem. It is essential that you have confidence in this part of the system.

Project planning helps, but you have to be careful here, too. Taken to its full potential, something like the computerised project management system Prince can be expensive and time-consuming to run. If you are not careful, everyone spends all their time in meetings and no real work gets done (a not uncommon public sector experience).

In the sci-fi comedy Red Dwarf, Rimmer showed that Zeno's Paradox - that motion is impossible because the moving object always has to cover half the distance - has a practical application.He spent three months planning his 12-months revision timetable for an exam, then revised the plan to fit the nine months remaining, and then revised it again, and so on until the day of the exam arrived without his eyes ever being cast on a page.

Of course, project management is vital.The public sector is littered with disasters caused by the absence of intelligent planning.

But projects with too much project management invariably run late. Project plans need to concentrate on the key milestones required for a project to succeed.

Organisations always have pressing day-to-day issues to tackle and usually cannot focus on implementation of the new system until the live date is very close. People will give it a quick look, but it is not until the moment of destiny is imminent that they seriously try to reconcile their existing knowledge of their organisation with the implications of the new system. Issues will always arise just before and just after a go-live.

Like Arthur Dent (Hitchhiker's Guide to the Galaxy), my motto is always 'do not panic'. But remember that innovation is often painful, and never perfect.

If you have done the basics right, do not lose confidence.There are always teething problems with any change.Take a few deep breaths. Sometimes, you just have to take the plunge.You will get the credit when the glitches end and the new system you planned is seen to work.

Markus Bolton is chief executive of a healthcare computer systems supplier.