The Scottish Government recently published their final report, “Benchmarking of Scottish Public Bodies’ Corporate Services 2010/11”. I will be quite frank and say that this report in its published form is almost entirely useless as it anonymises all the information and presents conclusions like:
The total cost of the ICT function ranged from £14,000 to £15.5 million in 2010/11. Costs of ICT were below £1 million in 21 organisations and were above £10 million in two public bodies. The organisations with higher costs often have specific ICT requirements because of the nature of their business, have a large number of users (employees and/or the public) and/or provide ICT services to other bodies. Since 2009/10, 21 organisations have had an increase in ICT costs and 17 have had a decrease. The differences level out when looking at the bodies as a whole, with average costs remaining largely the same in both years.[p. 3, 1]
Whilst this provides information, it provides practically nothing to highlight issues which may require investigation or provide opportunities for optimisation looking forward. However, the document goes on to state:
Going forward, the benchmarking information will be used by the central government sector in their contributions to the response to the McClelland Review of ICT Infrastructure in the Public Sector in Scotland.[p. 4, 1]
The McClelland report is all about making ICT more efficient in the public sector and particularly focuses on the need to share and combine services in order to obtain efficiencies of scale.
However, all is not lost. I contacted the Scottish Government and asked for the unpublished figures that were used to derive this report, particularly focusing on the ICT aspects of the report. They very efficiently sent me the spreadsheets along with cautions over the accuracy of the data. I have graphed some of the data to raise issues which are not highlighted by the benchmarking report but may be well worth investigating.
This graph shows the 38 out of 44 public bodies included in the report which provided information on the total cost of ICT per employee and the total cost of ICT. Each organisation is listed by its abbreviation a full index to these can be found in the Benchmark report[Appendix 1, 1]. Note that in March 2012 there were a total of 144 public bodies and “Participation amongst the remaining public bodies is encouraged, but is not mandatory”[p. 5, 1]. It seems highly surprising that this level of Management Information (MI) is not available and mandated from every public body, particularly as we are now over one year on from the McClelland report that specifically highlighted the poor availability of information regarding ICT costs.[p. 13, 2].
The x-axis of the graph shows the overall revenue of each listed organisation, so, for example the Scottish Government (SG) is listed as approximately £250m.
The y-axis of the graph shows the benchmarking calculation of the total ICT cost per employee within the organisation. In other words, as I understand it, this is how much it costs to provide a computer (usually 1 per employee) to an employee and provide the networking, software and support services to keep it operational for a year. This ranges from just over £500 per employee for the Scottish Prisons Service (SPS) to an amazing £13,000 per employee for the Registers of Scotland (ROS). Yes indeed, the figures do suggest that it costs just short of 25 times more to support a computer for a member of staff of the ROS than to the SPS! Indeed, it would appear that you could supply a member of staff of ROS with two brand new computers every month and still have change left over.
Looking at the graph it is hard to see why any of the organisations should be paying £2,000 to £13,000 per employee per year when there are a number of small public bodies that are able to provide the service for approximately £1,000 per employee per year. After all, at the top end you can buy a nice Apple computer for £1,000 which should last for at least 3 years, equalling £333 per year which leaves you with another £667 per year to pay for software, electricity, networking and support for one person. If every public body listed here were to run with these costs the figures suggest that we would save £55m per year, and note that this list does not include the NHS in Scotland or any of the local councils.
I suspect that some of these figures are misleading and probably include things that shouldn’t be included in employee costs like the costs of running public facing web sites. But that isn’t an explanation or excuse. At a time when experienced staff are loosing their jobs in order to save money, there can be no excuse for not being able to produce the figures that explain what is going on.
So, finally, in the graph the size of each dot represents the overall ICT expenditure of the organisation. So you can see that whilst the Office of the Scottish Charity Regulator (OSCR) has roughly the same cost of ICT per employee as the Scottish Qualifications Authority (SQA), the former’s absolute ICT expenditure is far smaller than the presumably larger SQA. Notable here is the fact that some of the organisations with the larger ICT expenditures have the highest costs per employee and some of those with the smallest expenditure have the lowest costs per user. This is potentially significant because there is a suggestion with the McClelland Report that costs can be saved by sharing services to obtain efficiencies of scale, but this graph would seem to suggest that scale does not guarantee efficiency.
I hope this article raises some important discussions and welcome critical feedback.
The Scottish Government, “Benchmarking of Scottish Public Bodies’ Corporate Services 2010/11”, June 2012, http://www.scotland.gov.uk/Publications/2012/06/1600.
John McClelland, “Review of ICT Infrastructure in the Public Sector in Scotland”, June 2011, http://www.scotland.gov.uk/Publications/2011/06/15104329/0.
Note that, in the interests of expediency and visual clarity the graph was produced using SVG graphics which will not work on versions of Internet Explorer prior to version 9 unless the Chrome Frame plugin has been installed. The graph should render fine on Chrome, Firefox, Safari and Webkit based mobile browsers such as those on iOS and Android devices. ↩