When low-balled projects go bad: Scottish pensions agency starts £10m procurement to buy the system Capita could not The Scottish Public Pensions Agency (SPPA) has begun tendering for a £10m administration system - rebooting a project previously awarded to Capita that was cancelled when the outsourcer missed all milestones. In a prior information notice, the agency – which administers about 500,000 people's pensions – said it is looking to " …
I'm always skeptical about any organisation whose "About Us" section has "Our Senior Leadership Team" listed first, above, you know, actually important things like "Our Purpose".
You just know that some narcissistic wanker "leader" demanded that.
Sadly, I'll bet there are far too many corporate websites which have the same problem. Still, at least they'll be first up against the wall, etc. (I'm not talking about the web developers, btw. Well, probably not.)
> "Capita was not able to provide a working system and did not achieve any of the project milestones. This was a main contributor to the project failure
> According to Audit Scotland, the SPPA spent £6.3m on the project and a further £2.4m extending contracts with existing suppliers when the project failed to meet the original timetable.
They spent £6.3 million on a project where Capita both failed to meet any milestones *and* failed to deliver a working system.
And they only got £700k back?
TBH, sounds like Capita were the winners here, especially since failures like this don't seem to have any impact on their ability to bid for future work...
> Sounds like poor contract drafting to me. No penalties for missing the milestones? No penalties for consequential costs (the £2.4M for the other suppliers)? Not even retention of payments due?
TBH, it sounds like Capita could have literally put zero people on the contract, defaulted on it and still come out of this with a large wodge of pure profit!
It's far worse than that.
"The project failure left the SPPA "unable to progress strategic, business and workforce plans as originally intended" and it therefore required an additional revenue budget of £9.8m between 2019/20 and 2022/23, as well as capital allocations of £13.6m "
So it's 6.3 + 2.4 + 9.8 + 13.6 for a grand total of £32.1 million. And Capital paid back £0.7m.
So yeah, get project, fail to deliver, profit from not having to pay back more than a tenth of what it got and impose four times the amount in cost to the organization that was foolish enough to trust it.
One of these days, the British Isles are going to get the message. One of these days.
Any time now, just you wait and see.
>Pensions legislation is some of the most complex there is. Good luck to anybody taking it on.
But is the same for everyone.
So unless there is something special about Scottish pensions that required some new breakthrough in hyper-mathematics, you just copy and paste the same system as every other council/company in the country
Not Invented Here.
They're probably taking a leaf out of the Downing Street playbook. It's got to be world beating. If it's world beating it can't be the same as the rest of the world is using so have to cobble up something new. Stir some overconfidence in there and they believe their own hype. The rest of us can see exactly where that's going.
But isn't your first question at the bid, " ever built one of these before?" if the answer is yes, then the next question is "did it work and can we have an exact copy?"
A city near here famously did this for property tax. They bought the software that another city ran and refused to change a line. They changed some bylaws to match the other city and put a PC in the line to the printer to replace the name of the town
"work" is a dubious concept if data that should be in parameters are hard-coded.
I have been in a similar position with a client. They were offering a service which processed data received via XML. I could see that one coming. The XML documents were going to be sufficiently large due to large numbers of work orders in each that they would bust the machine's memory if they were processed as a whole. A SAX parse would process the document an order at a time and throw it into the rest of the system with no problems but does need to have the element names hard-coded.
The solution was to run the incoming data through XSLT (Saxon can handle the file sizes to do that) and rewrite it into a dedicated namespace which is what the SAX parser used.
Next customer with a similar job but their own XML schema - no problem, just change the stylesheet. Give or take some minor tweaks (back-compatible) the core could remain common to multiple contracts.
Doing public tenders leaves very little room to reject a supplier because you simply don't trust them to actually deliver. This is part of the reason why companies like Capita keep getting projects like this. Whatever you say about them, they are really good at doing tenders, that is their main expertise.
I do see a potential way around this. If you where to keep track of the projects a company has done before you can determine a 'failure rate' and a percentage to go over budget on average. These two numbers can then be used in tenders, just apply the 'average over budget' factor before comparing prices and demand a failure rate below a certain percentage. That way you can take past performance into consideration, which generally doesn't happen right now. It also motivates companies to do a decent job because not doing so will harm their ability to win other tenders.
I like the idea but failure rate could be a bit tricky. Functions A, B & C were delivered but not D & E. Function G wasn't in the original but the client added it later and insisted that F was delayed to make way for it. All functions of different complexity and value. What's the failure rate?
500,000 pensions administered for £10m? Even if the term is only 5 years that's still only £4 per pension per year.
No doubt there will be a UK onshore requiements and possibly security clearances needed. Just the 3rd party costs will be more than this: hosting environment, software licences, identity and access management, annual audit. Then add the providers costs (assumed leveraged) for call centre people, telephony, NOC, SOC, servers etc etc.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
