I asked councils that filtered out traffic from internal IP addresses to provide data from both the filtered and unfiltered views.
There is debate about the value of filtering out traffic from internal IP addresses. Using internal IP addresses would commonly mean that the visit was from a member of staff, which is why some councils choose to filter them out. It is worth noting that it is not possible to tell if such a visit is being undertaken by someone in their role as a council employee or as a local resident.
Other people may visit from internal IP addresses, for example internal IP addresses might include public access PCs in libraries. It might also include professionals in schools who are customers of the education authority.
Similarly some internal users may visit the website from other IP addresses, for example because they are working from home.
Understanding what proportion of visits originates from internal IP addresses and whether these users behave differently would be helpful in understanding the impact of filtering out this traffic.
114 councils were able to provide details of visits to their main site with and without the filter. From these councils we can identify characteristics of the traffic originating from internal IP addresses.
Overall around 8% of visits in these authorities originate from internal IP addresses. County councils see the largest percentage of visits originating from internal IP addresses (a median rate of 12%). London boroughs see the smallest percentage of visits originating from internal IP addresses (a median rate of 5%).
Users originating from internal IP addresses behave differently and use different equipment compared to users overall.
Almost all visits from internal IP addresses are from desktop PCs.
Visitors using internal IP addresses are more likely to visit more than one page during a visit than visitors overall (a median bounce rate of 36.2%).
For most local authorities only a low proportion of visits originate from internal IP addresses. At whole site level it therefore probably makes little difference whether these visits are included or excluded.
That said, people visiting from internal IP addresses are much more likely to be using desktop PCs than visitors overall. They are also more likely to visit more than one page than visitors as a whole (lower bounce rate).
We cannot tell from these data whether they are more likely to visit some areas of the site than others.
Even without this detail there is hard evidence for something that digital teams often suspect: internal users are not typical of the vast majority of users of the website. Design decisions affecting the bulk of users should therefore not be largely based on feedback from internal users.
The visualisations used in this section are available as interactive tools on my Tableau site http://lawrd.co/1PestgJ.