Local authorities increasingly rely on web technology to provide information and deliver services to their citizens.

The majority of local authorities gather data on the use of their websites.

Very few of them publish these data. That is a pity because being able to compare website usage between authorities would be a useful tool for local authority web managers.

Though digital services (websites) should be primarily developed with and for the customers of the local authority, being able to understand how like (or unlike) your website is to others can be a powerful tool. At the very least it can prompt questions (‘why are we so different?’ ‘why are we not so different?’). Potentially it can be used to support business cases and to reality check assumptions (‘if we achieve this target we will be doing something only 10% of councils currently do’).

It might also be useful and interesting to people who manage websites in other sectors.

In a local community, a local council provides an interesting benchmark. It is likely to be a high volume digital service used by a high proportion of the local residents. A range of other stakeholders, such as businesses, voluntary organisations and colleges, could find data about the usage of the local authority’s website helpful in business planning and reality checking their own analytics data.

This study set out to gather some key data about website usage and to demonstrate the value to the sector of sharing these data.

Data collection

I wrote to all local authorities in the United Kingdom requesting data about the usage of their website. The text of the request is attached as Appendix 1. I used the email addresses held on the excellent FOI Directory site www.foi.directory.

I did not attempt to define what I meant by ‘their website’ and this did not seem to cause any significant problems. Some councils contacted me to clarify that they managed a large number of properties and to ask if I wanted data for all of them.

The data I requested was for the 12 months ending 31 July 2015.

The report covers data from principal local authorities (ie not parish or community councils) for England, Scotland and Wales. It does not include dedicated sites for fire and rescue authorities or combined authorities.

It became clear that the recent reorganisation of local government in Northern Ireland has complicated the collection of website usage data over that twelve month period. It also complicated identifying the correct address for FOIA requests and so, unfortunately, Northern Ireland data is not included in this analysis.

I wrote to 408 addresses in batches between 13 August and 18 August 2015. All replies should have been received by 18 September and I began the analysis on 9 October 2015. By that point 11 councils had not provided a response. 2 councils responded to say they refused to accept FOIA requests by email and 2 councils responded to say they did not hold the data requested.

I was cognisant of the fact that making a widespread FOIA request like this involves hundreds of public sector workers in pulling together the data. Accordingly I tried to minimise the amount of work that would need to be undertaken. I asked for a small number of data points and picked data that I knew to be readily available via default reports in the widely used Google Analytics. This necessarily reduced the level of detail that is available in this report as I tried to strike a balance between not wishing to create unnecessary work across local government while still being able to provide useful insights.

I am very grateful to the many people who were involved in providing this data, many of whom included additional notes or observations.

Slightly belatedly I wrote a blog post explaining what I intended to do with the data http://benproctor.co.uk/am-i-normal/ and this received a very positive response from digital professionals in the sector.

Data entry

The vast majority of responses were returned via email. A very small number of councils actively publish these data and most of them helpfully extracted the datapoints I had requested.

Where authorities provided data for multiple sites I entered only the data for the site with the highest traffic (almost always www.thecouncil.gov.uk) and additional sites with at least 20% of the traffic. The vast majority of these additional sites had levels of visits that were vanishingly small compared to the site with the highest traffic.

I entered data for filtered and unfiltered views where they were available, and for mobile sites where they were available. In fact only 6 dedicated mobile sites were provided.

The data were entered into a Google Spreadsheet. It is possible that I may have made errors in data entry and I am publishing the full spreadsheet partly to enable people to identify such errors. I also consider it good practice to provide the data underpinning this analysis.

Data used in the analysis


A ‘visit’ (or ‘session’) occurs when a user requests a page or otherwise triggers an event on a website. Further events or pages are classed as occurring within the same visit unless the gap between them exceeds a preset limit. The default limit in Google Analytics is 30 minutes. I did not ask local authorities what limit they applied to their data.

‘Maximum data’ includes unfiltered (where available: filtered otherwise) and all other data available, including from councils who were not able to provide 12 months of data.

‘Best available data’ uses filtered where available, unfiltered otherwise, and excludes data from councils who were not able to provide 12 months of data.

The analysis of visits from internal IP addresses only includes councils who provided both unfiltered and filtered, but does include data from councils who were not able to provide 12 months of data.

Bounce rates

A ‘bounce’ occurs if a user requests a single page (or triggers a single event) and then does not visit another page (or trigger another event) within the time limit set. In Google Analytics the default time period is 30 minutes. I did not ask local authorities what limit they applied to their data.

Where bounce rate was provided as a percentage or a proportion, this figure was used directly. If the number of bounced sessions was provided, this was used to calculate the bounce rate.

‘Best available data’ uses filtered where available, unfiltered otherwise.

Filtered and unfiltered are treated separately for the section on internal users.

User data: visits from desktop, mobile and tablet

Analytics tools usually record data about the operating system and the browser used for the visit. Google Analytics and some versions of some other tools automatically classify these as ‘desktop, mobile or tablet.’ These automatic classifications have been used, but where local authorities only provided a breakdown of operating system and browser these data have not been used.

‘Best available data for desktop and tablet’ uses filtered data where available, unfiltered where it is not. ‘Mobile’ uses data from mobile sites where available, filtered where not and unfiltered where this is all that was available.

Other statistics about local authorities

Population figures are 2014 estimates from the Office for National Statistics (ONS).

Population density figures are 2012 estimates from ONS.

Deprivation figures are based on the 2015 release of the indices of multiple deprivation (IMD). They apply only to England. The figures used here are the average IMD rank per local authority. Each LSOA (Lower Super Output Area, a small geographical area typically smaller than a council ward) is ranked against the others. Rank 1 is the most deprived. For each local authority the average rank of all the LSOAs in the local authority area is taken. The larger the average, the less deprived the local authority.

This means, inevitably, that it is not clear whether any observations drawn from analysis involving deprivation data will hold true for Wales or Scotland. To do so would rely on an assumption that Wales and Scotland were sufficiently similar.

Re-use of data

Despite the fact that I was requesting data, a small number of councils applied restrictive licenses in their responses. I assume these were default conditions attached without considering the nature of this response but until the situation has been clarified anyone seeking to re-use the data should be aware of this.

<< Back to the index

Visits >>