Frameworks for government are important. They create a forum in which companies can set out their pitch for what they can do for government. They allow the government to undertake a certain amount of vetting of those companies. And they reduce the bureaucracy associated with government’s procurement.
Necessarily, their scope is limited. The scope of the framework is clearly defined. Companies specify their competencies against that scope, and this allows the public sector to buy services within that scope via the framework. It all works rather nicely.
However, risk is introduced when that scope is stretched.
Let’s say a framework is geared around a specific subject matter, cloud computing, for example. And let’s say that in being accepted onto the framework, a company demonstrates its competence in this area.
Now let’s say that a government organisation seeks expertise in a different subject matter, the provision of school meals for example. And let’s say it looks on the framework, and decides to go with the aforementioned company to provide its school meals, predominantly because it seems to offer good value for money.
There are two issues here. First, if the company is not very good at providing school meals, where do we go? The framework offered a level of comfort and protection. The companies that are on it have shown a level of competence in cloud computing. But they have shown no such ability in cookery.
Second, what if there’s another company that’s top dog in the school meals world. They make school meals that kids rave about, with locally-sourced ingredients at a price that has to be seen to be believed.
They see the business being awarded within the cloud computing framework. And, rightly, they say “hang on a minute”. This second company had read the terms of the cloud computing framework and dismissed it outright. “We don’t do cloud computing. We’re good at cooking food for kids. How could we possibly benefit from being listed on there?”
The example above is an extreme one for illustrative purposes. (Funnily enough, this specific example was presented to me earlier today by a Cabinet Office employee. They suggested that if the above deal represented good value for money, what was the issue?)
What are your thoughts? Should frameworks be stretched beyond their initial intent and their advertised scope to enable potential savings to the taxpayer? Or should their scope be carefully policed to ensure that companies are not awarded business unfairly at the expense of others; and that government doesn’t do business with companies ill-equipped to meet their requirements?
Yesterday, G-Cloud celebrated its third birthday.
In that time, it has accounted for £431m of government expenditure. (The correct figure is allegedly £467m, but the team is yet to update its data file to correct a bug I highlighted to them, so I can’t vouch for the latter figure. They’re hoping to correct this today.)
Of the £431m, £50m (12%) has been spent on software as a service (SaaS); £31m (7%) on infrastructure as a service (IaaS); £6m (1%) on platform as a service (PaaS); and a staggering £344m (80%) on “specialist cloud services”, defined as follows:
Specialist Cloud Services (SCS) support your transition to SaaS, PaaS and IaaS. Examples of SCS include cloud strategy, data transfer between providers or day-to-day support of cloud-based services.
My take on this is that the SaaS, IaaS and PaaS are on the out-of-the-box widgets themselves; and “specialist cloud services” are the people who are needed to either implement or support cloud.
But if you were to buy a truly cloud-based service, one that plugged and played, then even though there are humans needed to support this on a day-to-day basis, my view is that their effort and cost would be wrapped up into the *aaS cost, and would be labelled such. For example, every penny of the £2,970,196 that Ninian Solutions (Huddle’s trading name) has earned through G-Cloud is labelled as SaaS. If you want Huddle, you sign a contract; you get some licences; and it’s yours to use. There are humans who support it on a day-to-day basis – hell, I’ve met lots of them. But to the customer, these people are invisible.
So I expect that the humans covered by the SCS bucket above are doing the stuff that *doesn’t* come out of the box – either migrating stuff to the new solution, or doing something cloud-related that doesn’t come out of the proverbial box. One hopes that they’re not doing stuff unrelated to cloud. That wouldn’t be good, would it?
In my rather simplistic mind, cloud has two features. First, it’s available solely over the internet. And secondly, it’s shared. If it’s not available over the internet, then it becomes infrastructure sat in a semi-dedicated data centre, which is the old world of IT, and therefore doesn’t fit the cloud definition. It’s available over a network, but that network is a pipe dedicated for use by the end client.
And if it’s an application that is designed and/or implemented specifically for your use, then it doesn’t embrace the ethos of cloud that is associated with driving down cost through reuse.
I would have hoped that by embracing the two features above, the extent to which humans was needed would be much less than the above numbers suggest to be the case.
There is clearly a need for people to implement the things that you are buying through G-Cloud. But I am surprised that 80% of the money that is being spent on G-Cloud is dedicated to this. I have no idea what an appropriate percentage would be – indeed it might be entirely appropriate for four pounds in every five to be spent on the people associated with the service. Views?
G-Cloud is a fabulous thing. It has transformed the way in which certain elements of IT are bought within government, and has brought into the fold a large number of small- to medium-sized supplier organisations. But I find certain elements troubling.
Valtech, BJSS, Methods Advisory, Equal Experts, PA Consulting and IBM UK are the biggest six suppliers based on total evidenced spend, with a total take of £88m. Of that, £87,188 has been spent on *aaS. That’s less than 0.1%. All of the remainder has been spent on the humans to think about, implement and support this stuff. I wonder whether that was the model intended when G-Cloud was introduced.
At the end of January, my 3Gb giffgaff goody-bag expired. I use it to access the internet from my BYOD via a dongle at one of my clients. I don’t set it to top up automatically, as often, I don’t visit said client for a few days after its expiry.
So I went to the giffgaff site to renew the expired £12 goody-bag. But when doing so, I noticed an £18 goody-bag that had unlimited internet. Wowzers! (It also came with 2,000 free minutes of calls, but that’s of little use to dongle-using Dan. Although as you’ll see, it’s relevant to the story.) I added it to my basket and bought. Happy days!
A few days into my new goody-bag’s life, I received a warning message from giffgaff saying that I was not allowed to tether on an unlimited data plan, and that my internet access had been halted until such time that the tethering stopped. (Neat that they could tell.) I was mildly irritated at that specific condition not having been actively presented to me upon buying (I’d had no idea), but accepted their judgment and tried to go about rectifying the issue.
I posted a comment on their forum asking how I might downgrade immediately to the 3Gb goody-bag. Ah, it seems I can’t. You can only buy a goody-bag if the previous goody-bag has either (a) expired; (b) run out of data; or (c) run out of minutes.
Hmmm. The unlimited goody-bag wasn’t due to expire until 28 February, so that wouldn’t do. I needed it before then. It had unlimited data, so it was unlikely that it’d run out of data. Ever. So I was left with one course of action: drain the minutes.
Harder than you might think.
I needed an old phone, one that could accept the old-style chunky giffgaff Sim card. The only one I had was an old iPhone that literally would no longer accept any charge. So my good friend Bal hand-delivered a cutting-edge Sony Ericsson W995 to my door this morning.
After giving it sufficient charge for it to function, I discovered that it was locked to the Orange network. Expletives abounded. So I took it to a shop to get it unlocked for £12. (Hope that’s OK, Bal.)
And now the phone is sat downstairs connected to its charger. At 8:09pm, it made a call to my main mobile. That call is currently two hours and four minutes old, and counting. To make sure the call doesn’t get cut off (I’ve heard that silence on both ends of a phone line result in it being cut off), I’ve positioned it in front of a speaker playing a Spotify playlist on repeat. (As I type, it’s piping Lionel Richie’s Hello into my main phone.)
My plan is to continue the call to myself overnight tonight, and ideally throughout tomorrow. Assuming no interruptions, my 2,000 minutes (33 hours and 20 minutes in old money) will expire at 5.29am on Monday morning, at which time I’ll be able to buy a 3Gb goody-bag.