Tag Archives: Data

Final, cumulative Digital Britain web analytics

We recently discussed how by using Google Analytics and a time stamped RSS feed, we could get a good idea of when people were reading and commenting on the Digital Britain – Interim Report. Now that the official consultation period has ended, we can show you the full web analytics from our re-publication of the report and try to draw some conclusions. We’d appreciate any insight you might be able to bring to this and have made a fairly complete set of PDF exports of the analytics available on Scribd. We’re aware that web analytics are never 100% accurate, but these should be accurate enought to indicate trends. Below are some images which offer a useful overview. Click on them to go to Scribd.

Visitors Overview

Visitors Overview

So 3,313 people made 4,409 visits and on average viewed 2.75 pages over 3:03 minutes. About half of these visitors, ‘bounced’; that is, they came to the site and left without reading beyond the initial page. The spikes in traffic are when we launched and when the Guardian, boingboing and the Digital Britain discussion site linked to us.

Traffic Sources Overview

Traffic Sources Overview

Note how few people came to the site via a search engine. Currently, a search for ‘digital britain report’ on http://google.co.uk, ranks our re-publication of the report at site #15 (the Fake Digital Britain Report ranks slightly higher).



When people did come to us via a search engine, the searches are predictable, they stayed on the site for less than the average time and tended to ‘bounce’ more often than average, too. The site is optimised fairly well for Search Engines (there are areas for improvement) but perhaps because it’s a new site, with few incoming links and is competing with established government and newspaper sites, it would be difficult to rank higher in this short period of time.

All Traffic Sources

All Traffic Sources

The top ten traffic sources accounted for 83% of all of our traffic. 43% came from Twitter and ‘direct’ sources which I suspect are the shortened URLs which are used on Twitter. I think we can say that without Twitter, we would have had much less impact and found it more difficult to advocate comment on the report given the time we have available to work on WriteToReply.

Content Viewed by Title

Content Viewed by Title

With just over 12,000 page views, the front page of the report is by far the most popular page but it’s interesting to see that the 3rd and 4th most popular pages are the ‘Comments by User’ and ‘Comments by Section’ pages, which are unique (??) to CommentPress, the tool that enables paragraph level commenting.

UK geographical coverage

UK geographical coverage

We had visitors from across the UK, but the largest concentration by far came from London (44%). Next was Manchester (5%) and third was Milton Keynes (2.6%). I should note that Tony (one half of WriteToReply) works for the Open University which is based in Milton Keynes. The rest of the geographical coverage would appear to be just small clusters of people, mainly in England and the urban areas of Scotland and Wales. If WriteToReply is to be a successful public service, we need to see this coverage spread more evenly throughout the UK for national consultations. We’d also like to see local authorities and local interest groups use WriteToReply for their consultations, too and their use and advocacy of it might also help even the spread during national consultations.

Connection Speeds

Connection Speeds

At least 70% of visitors came to the report over a ‘broadband’ type connection. This is possibly much higher if the ‘unknown’ speeds were also mostly broadband, too.  Interestingly, at least 17% of visits were from a T1 connection, which is a leased line used by business. Just 2% of visitors were confirmed dial-up users.

Browsers and OS

Browsers and OS

54% of visitors browsed using Firefox, while 26% used Internet Explorer. 66% ran on Windows and 23% ran on a Mac.

So, if you run Firefox on Windows, live in London with a broadband connection, came to the site via Twitter, spent about three minutes on the site (not long enough to comment) and never returned to the site… you’re pretty average, but we thank you for your interest. 🙂 If you’re one of the 69 people that left some of the 295 comments on the report via WriteToReply, we want to thank you very much.

And do remember that anyone re-publishing a document on WriteToReply can easily collect this kind of data to give a better understanding about the impact the consultation is having. Were it endorsed and used by civil servants to publish consultations, a much more accurate overall picture of who is engaging with the consultation and how they engage could be built up and why not make that information publicly available, too?

Clouds of Comments

A couple of days ago, we noticed that the Digital Britain report had received over 100 comments. So to celebrate I created a Wordle word cloud from the RSS feed of comments. The feed currently provides up to 200 comments, so I was able to pull everything into Wordle. We mentioned it on Twitter, linking to the image on WriteToReply’s Flickr account and that made some people happy.

I know what you’re thinking: Wordle’s yesterday’s toy. You can’t turn around without seeing a bloody Wordle these days. And you’re right. In the back of my mind, I knew we could do more to celebrate passing ‘the centenary’. So did Tony…

So duty-bound, I looked around and found FeedVis. It’s free software and a really slick way of getting an overview of the comment trends, browsing the comments and navigating through to the full comments on the report. Don’t tell me it’s yesterday’s toy. Please.

It’s interesting to compare the Wordle word cloud with the word cloud produced by FeedVis. ‘Music’ and ‘Industry’ feature heavily in the first cloud, but less so in the second, where ‘access’, ‘content’, ‘digital’, industry’ and ‘technological’ are all favoured pretty equally in terms of how often they’re repeated. By default, Wordle looks for a maximum of the most popular 150 words and shows their relative popularity by increasing or decreasing the size of the word. FeedVis is more sophisticated. It uses stemming (where Wordle doesn’t), so words like ‘fast’ and ‘faster’ (<cough> broadband <cough>) are mapped for the purpose of counting their frequency. There’s a couple of other important features, too:

  • The first is frequency. Frequency says how many times a word is used per 1000 words. If you hover over a word, you’ll see its frequency to the left of the frequency change value.
  • The second is frequency change. Often, a word will be more (or less) popular than usual in a certain time period (for instance, “election” in early November). Frequency change measures that difference as a percentage: greener words are unusually popular; redder words are the opposite.

Here’s a screenshot from earlier today. Click on the image to visit the FeedVis of the Digital Britain – Interim Report itself. Hopefully, we can find more tools like this to add to the WriteToReply suite of ways which comments can be read, viewed and interpreted by us all, including the authors of the reports. If you know of any, please leave a comment. Thanks.

Digital Britain: First week analytics

In the interests of transparency and fun, here’s an overview of the Google Analytics from the first week of Write to Reply’s re-publication of the Digital Britain – Interim Report. Note that GA has certain limitations that can make the reports a little misleading, but heck, they’re the best we have.

As we explained here, Tony and I threw the site together over a couple of evenings and launched it over Twitter. We’ve continued to watch Twitter as the main source of feedback about Write to Reply. A running Twitter search (via TweetDeck in my case) for ‘digitalbritain OR “Digital Britain” OR writetoreply OR “Write to Reply”‘ quickly alerts us to any possible reference on Twitter to either the site or the report. (This coverage extends somewhat to the blogosphere, also, which we’re also tracking in terms of things like incoming links to the site.)

When we see someone referring to the Digital Britain report, we might contact them (@ them on Twitter, or post a comment to a blog post) and mention Write To Reply, where the report is commentable upon.  Every response I’ve had to this approach has been appreciative.

Many people have blogged about Write to Reply’s re-publication of the Digital Britain report but two posts have generated a significant amount of traffic. One by Jack Schofield on the Guardian Tech blog and another by Cory Doctorow on boingboing. Both writers urged their readers to use Write to Reply to comment on the report.  Thanks!

So first up, here’s a graph produced by the Akismet spam filter used on Write to Reply. It’s not entirely accurate for some reason (!?) but we can confirm there’s been three spam comments and 87 genuine comments received.  Thank you!

Given what I’ve said above about our use of Twitter, the first of the GA reports is not too surprising (click on the images to see them full-size and legible):

This shows that overwhelmingly, our traffic is via Twitter and, most likely, shortened URLs used on Twitter (which accounts for the ‘direct’ traffic). Thanks Tweeple, for about half of the traffic to the report!

The Guardian and boingboing are, respectively, the second and third largest referers and as they are busy news sites, we’ve probably seen most of the traffic they will generate as people turn on to the latest news. Maybe I’m wrong there. We’ll see next week.

When people visit the report, they’re mostly landing on the front page, as we would expect. Second on the list (with a much lower landing rate) is Section 5.1 Education and Skills (a section that was linked to directly with a request for comments in a message that got retweeted several times on Twitter).

48.15% of people are clicking through to at least one more page (actually, the average page views is 2.56/visit) and the top of these exit pages is, again, Section 5.1 Education and Skills.

Want to know about aggregated page views? As of 9pm last night, there had been 5683 views of 54 pages in the last week.

As you can see, besides the front pages of the report, people seem to like looking at the list of comments by section and the list of comment by user.  This suggests that that specific functionality provided by the CommentPress theme for WordPress is of interest to people. Excellent! 🙂

Most visitors have visited just once, but 19% are returning visitors (and we commend your dedication!).

So, where are all these people from? Mostly the UK (81%) but also from the USA (8%) and 40 other countries.

People across the UK are visiting the Digital Britain report, though almost half (48%) of visitors are from London (does anyone know exactly how Google locates people? Is it locating actual people at computers or just their ISPs?)

Interestingly, 46 visitors are from Lambeth, which is just minutes away from Westminster. We can only hope that these visitors are a concentration of people working in government 😉

Finally, here’s the big picture. The site peaked on the first day (Wednesday 4th Feb), dropped rapidly but slowed slightly due to the Guardian tech blog post (Thursday 5th Feb), continued dropping over the weekend until Monday 9th when it picked up again and then rose again suddenly, thanks to boingboing’s post on Tuesday 10th.

If anyone wants view access to the full analytics, send us your GMail address and we’ll add you to the user list. We’ll post more on this subject on March 13th, after the consultation period ends or if anything interesting happens before then. If you’ve got ideas about how this data can be creatively and productively used, please get in touch. We have some ideas on the burner, but welcome yours too. Thanks.