Ok, Tuesday was quite a busy day for me and many of my colleagues here at The National Archives. It was brilliant to launch Operation War Diary and provide access to the unit war diaries online. If you haven’t already, you really must take a look at our First World War collections, and preferably get tagging!
For the first time we were able to make use of real time web analytics. This means that we could see data about how people were using our website second by second – where people were coming from, what pages were being used and what sorts of devices people were using.
A couple of things really stood out during the day and I wanted to share these with you. This is by no means a comprehensive evaluation of yesterday’s activity, just some observations by someone who loves statistics…
Mobile is more popular than desktop at some parts of the day
When I blogged about the redesign of nationalarchives.gov.uk back in September, I explained that one of the reasons for taking a new approach was that almost a fifth of our audience was using a tablet or mobile. But this is basically a flat average, and takes no account of daily fluctuation. Tracking the data in real time yesterday changed my view of that completely as it showed how the devices you use change at different times of the day.
During the working day, when you’re fairly desk-bound, desktop PCs rule – the highest peak I saw was at 1pm when 88% visitors (12% mobile/tablet) accessing www.nationalarchives.gov.uk were using a desktop. I’m *guessing* that 1pm was a desktop peak because so many of us eat lunch at our desks these days. But 25 minutes later when the BBC1 news had run a piece, desktop usage had dropped to 63% with mobile/tablet rising to 37%. This is really strong evidence of second screening – where viewers are using one device to watch TV, and a second device to enhance their experience, or interact with what they are watching. In the afternoon, desktop usage stepped back up again.
This second screening trend showed more strongly later too. During the same feature on the news at 6pm on BBC1, tablets/mobiles overtook desktops for the first time that I had seen – 55% visitors were using their mobile devices. Then during the 10pm news on BBC1, the split peaked at 59% mobile/tablet to 40% desktop (in fact, tablets were the most popular device type – 41%).
Web analytics for 15 January 2014, 10:30pm
Mobile/tablet usage also rose when we’re commuting (hopefully on the train or bus, rather than the car…) and in the evening – from about 15% throughout the working afternoon, up to 30% on the evening commute as we caught up on what’s going on in the world.
We love TV and our behaviour is influenced by it
Looking at the trend throughout the day it’s really clear that our traffic peaks coincided with TV features. As well as the switch in devices used to access our site, a few other trends stood out as well when we’re gaining a lot of TV coverage:
- UK usage vastly increased. For example, our UK audience during our feature on the BBC1 evening news was 93% at 6.25pm (25 minutes earlier, it was 79%)
- We gain lots of new visitors. 83% of visitors were new during the 6pm news feature, compared to 71% at the start of the news
So yes, we’re a nation of TV lovers!
So why does all of this matter?
To some extent, it’s no surprise, but I was taken aback at the sheer fluctuation. It’s a real lesson in the importance of drilling down deep into data, and not just relying on averages. Likewise, I’m not going to rely purely on a day’s worth of peak data – I’m really interested to know what a typical day is like, and the differences between working weeks and weekends – but it’s really opened my eyes.
It matters because it will help us design better services – and I mean design in the broadest sense – layout, visuals, language, function. If we thought that mobile was important back in September, then Tuesday has completely reinforced that for me and should transform how we design all our services.
Hi Emma, yea stats OK but its all very subjective. Be good to see a weekend trend and I note your use of the term “busy”. You were just “quite” busy not “very busy” then? Can you measure it at all? As long as you are never “too” busy
Ian
Hi Ian,
The stats are a fairly evidence-based approach – I didn’t want to rely on anecdote or supposition. It seemed reasonable to me to correlate TV coverage (which I was watching) with an increase in traffic to the website (which I was watching) and a significant change in device. The fact that the same pattern occurred three times (1pm news, 6pm news and 10pm news) made me think that this wasn’t a coincidence. With the lunchtime/desk-based comment, I did say that I was guessing. If you can draw different conclusions to what I’ve suggested, let me know – the more ideas the merrier – and it helps us to build up a picture of what our visitors need. This has piqued my interest and I intend to investigate realtime traffic patterns on a “normal” weekday and a “normal” weekend and I will write about this once I’ve done more research. To me one of the joys of statistics is that they can be interpreted in different ways. Thank you for your interest.
Emma
Interesting stuff. The use and abuse of web usage stats is well documented and as you say, the main information we can gauge from stats such as these are patterns in the data. I had not really considered before how people switch between devices over the course of the day, though I am now very aware that I do this myself – mostly in the pattern you describe. The only difference of course is that those of us who work in archives may not be able to eat our lunch at our desks therefore would switch from a PC to a mobile device at approximately 12.30!
Hi Jenny,
I hadn’t considered the archivist/lunch perspective! Although eating at our desks is not a good habit, it’s nice for us non-archivists to have the choice. At some points on Tuesday, I found myself third screening…
Emma