Novopay a $30m horror show

20:52, Nov 25 2012
Bloody computers
GIVE 'EM THE REBOOT: Behind every computer mess is usually a heap of user-error.

Novopay is a horror show. A $30 million school payroll system due for completion in 2010 is finally rolled out in August 2012 with a price-tag closer to $100m. Thousands of staff are underpaid, overpaid or not paid at all.

Next, the company that built the flaw-ridden system insists it's "rock solid" and blames klutzy school staff for entering data wrongly. Meanwhile, journalists uncover evidence that bad omens were ignored right up to the "go-live" date, and small-scale trials were abandoned in the panic to meet deadlines.

It is, says Otago University's Professor Robin Gauld, all too familiar. For decades, our government has been blowing hundreds of millions of dollars on useless IT systems. What's worse, it's almost inevitable it will keep on doing so.

In 2006, Gauld co-authored an influential book, Dangerous Enthusiasms, which sought to explain why public sector IT projects so often crash and burn.

The book's conclusions are dispiriting. Quite simply, large-scale IT projects almost always fail, no matter who's doing them - deadlines are missed, budgets are blown, promised functions don't work. Projects are abandoned at huge cost without ever being used.

Business history is littered with pricey IT cock-ups (American candy-maker Hershey recently lost US$150m in sales because of the rushed rollout of an inventory system), but the biggest disasters of all tend to be in the public sector. A 2001 US study found the success rate (ie, on time, on budget, works) for IT projects was around 59 per cent in the retail sector, 27 per cent in manufacturing and 18 per cent in government. Some 28 per cent were abandoned, or failed altogether.


Worse still, the bigger the budget, the worse the odds of success: the same study showed projects worth less than US$750,000 have a 55 per cent success rate, while not a single project of more than US$10m was fully successful.

And although Novopay may look horrendous, and will no doubt ruin Christmas for numerous teachers, at least it's functioning to some extent. New Zealand's most spectacular IT cock-ups have gone up in flames before takeoff.

Take INCIS, the police's "Integrated National Crime Investigation System", developed for 10 years at a cost of $100m and finally abandoned in 1999 having never become operational. Or the $17m spent on a Health Waikato computer system that was abandoned in 2000, or the $26m spent on a similar system for Capital Coast Health that only ever performed a fraction of its intended functions.

Abroad, where populations and budgets are so much bigger, the price of failure is eye-watering. In the UK, £20 billion has been sunk in a scandal-ridden and uncompleted project to computerise the National Health Service. In Australia, a botched health payroll system built by IBM for Queensland Health has cost A$500m, and has seen 50,000 staff being overpaid more than A$90m.

So what's going wrong? Aren't computer geeks meant to be smart?

In his book, Gauld and co-author Shaun Goldfinch concluded that big IT projects fail because of four "dangerous enthusiasms".

First you have enthusiastic and ambitious salespeople, who will say anything to get their foot in the door.

Then you have managers with performance-related contracts (particularly prevalent in New Zealand's deregulated public sector) determined to make their mark. "So they're very keen to get new wizardry in place and to forge deals and show they've made a difference."

Add to that "technophilia" - the wild-eyed belief that technology is intrinsically marvellous - and "managerial faddism", and you've got the four-part recipe for calamity. Behind all of those reasons, of course, is the big one: putting together a big IT project is really hard.

"These projects are very complicated," says Gauld. "People often have huge visions and promises are made, but they're not able to be delivered on. It's a bit like building a grand mansion that looks great on plan, but the side of the hill starts to slump and it costs more money to fix before you can build five floors on it."

People also change their minds, says Gauld.

"It's just like building a house. You realise you want the kitchen a bit different. That wasn't in the original contract, so you pay for it. It's the same with IT. As things evolve you might see new functions you might want, so you pay for them."

But even if the buyer seems to be getting fussy, most fault ultimately lies with the vendor, says Gauld.

"They need to be really realistic in terms of what they have to deliver. They have sales to make, but they have an ethical obligation to be realistic." Outright corruption is seldom part of the IT-failure equation, "but there's a fine line between what you'd call corruption and overzealous salesmanship".

LAST WEEK, Talent2, the Australian company that designed Novopay, finally spoke publicly about the debacle, and said its technology was "rock solid", and that the problems were to do with human error.

As insulting as many school staff found this, there is probably a grain of truth there.

Funnily enough, Talent2 designed the payroll system of Gauld's employer, Otago University, and Gauld says that system works just fine. Five thousand-odd people get paid on time, despite complex shifts and pay-rates and leave arrangements.

Novopay, however, manages the pay of around 100,000 people spread over thousands of schools, each of which is technically a different employer.

In IT, size does matter, and it pays to test things on a small scale before going the whole hog.

Which makes it all the more mysterious that Talent2 and the Ministry of Education skipped the vital failsafe of running a single-region pilot of Novopay.

Last week, the Star-Times reported that the ministry dumped a planned regional rollout because it was desperate to meet self-imposed deadlines. And today we report that the ministry ignored an internal survey showing only a third of Novopay users thought the system was ready.

Gauld says such rush-jobs are all too common. Indeed he believes the National Government may be setting the stage for future horrors with its stated policy of getting a national patient-record system in place by 2014.

Another Novopay-style cock-up may occur unless the planned health IT system is thoroughly tested on a small scale, but with 2014 looming, Gauld is pessimistic.

"I don't think it's going to be possible to get it right, and then there's going to be a national rollout at a great cost."

Perhaps the biggest mistake in the Novopay saga is that no one had the willpower to stand up and say "no we're not ready - we're not going to do this".

Around the country, New Zealanders are paying for that failure of willpower.

The competition is stiff, but in the pantheon of New Zealand computer cock-ups, the Novopay debacle is "up there", says Gauld.

"It's definitely a failure if you think of failure as being over budget and not on time. And if you're one of the victims of their system, who's not been paid - for those individuals it's a grand failure."

Sunday Star Times