I finally had some time to delve into recent statistics for the site. Apart from the usual election year rise of the readership, the demographics have been moving a lot.
The most noticeable statistic in the recent months has been to do with gender. Google Analytics has an limited ability to match the stated gender of some of our users. Typically this is between 40% and 65% of our readers. Not particularly accurate, but certainly of interest in how it changes.
As part of my normal duties to keeping an eye on the site health, I normally tend to watch the percentage of female readers every few weeks. For me it serves as a pretty good surrogate for how inclusive and accessible this site is. When it drops I start hunting for dirty socks and smelly jockstraps who have started to infest the place. But in recent months I have been somewhat remiss because of excessive work hours. As a percentage of sessions it seems to normally live at around 25% to 30% and fluctuates according to the types of stories and depth of the locker room foolishness in the comments section. I am not sure what I would have to look forward to if the site ever started to reach gender parity.
I’ve grabbed some of the weeks from earlier in the year to give context, and sampled progressively more in the weeks heading up into August and September. Because of the variability of the sample population, I’ve only put in the percentages of the sampled population to make it easier to see the trends. And I like to read numbers, so I am a bit lazy to make graphs today. I have highlighted the periods of interest and put my interpretation down below the table.
|Week starting||% Sessions||% New|
We have increased the percentage of women readers from about a quarter earlier in the year to just under 40% in recent weeks. The majority of new readers (people that haven’t seen the site previously) are now women.
Now this kind of interest swing isn’t abnormal. But I’d have to say that in the 3 years since this facility has become available, I’ve only seen seen our readership of women rise above 34% of the sessions once. In the available data that was in the election week in 2014 when it hit 42%
It appears to have started with Metiria Turei’s revelations about dealing with WINZ back in July. We had a sudden spike in interest from what looks like mostly the existing women readers of the site in the days.
Surprisingly, to me at least, the site didn’t get a big bump in the readership of women when Jacinda Arden became leader of parliamentary Labour in early August. I suspect that the volume of interest on that day was more normally spread.
What we did get was a strong spike of new women reader interest shortly after Metiria Turei announced her resignation from the Greens leadership and in the subsequent social media storm over that. The spike at 60.5% of new readers, mostly in one day, was only matched by election week in 2014.
The last few weeks of the election campaign has been seen a steady rise in the readership of women. A lot more new women readers with more arriving all of the time. The last few weeks have almost put the site at election 2014 level already and is far higher than it was in 2014 when it was 33% and 29% of sessions in the same two weeks before the final week. It will be interesting seeing what this week does and if it is higher then the 42% of 2014.
But I suspect that we have far more interest in this election by women than we did in 2014.
The generational stats are also showing quite a lot of change since 2014. I’ll look at them later in the week.
This site: Just under 90% of this site’s readers are from New Zealand and most of the overseas visitors come from places that kiwis like to frequent, mostly Aussie, US and the UK. This seldom changes very much. In the last 30 days we have more than 55 thousand unique users. This year to date more than 386 thousand unique users have been on the site.
Users: Google has a pretty good grasp on who different people are. This is based largely on google cookies being set by web sites like ours and net utilities like facebook and gmail. Since cookies are used for all types of purposes like keeping track of logins, and filling in fields few people try to evade them. These are the unique users that google can identify. It will be slightly inaccurate because sometimes it cannot detect of the same person is accessing from different devices. But it is pretty accurate at detecting unique users.
Bounce Rate: Effectively this is the percentage of readers who read a single page and who jump away virtually immediately. This is usually low on this site but jumps sharply with social media and net search spikes.
Sessions: This is a pretty arbitrary measurement that attempts to figure out how long someone remains glued to a website. It typically has a timeout of no activity in the order of a number of minutes. Used in conjunction with % New Sessions, New Users, Pages per Session and Average Session Duration, it gives a good indications of where a surge in readers is coming from.
For instance, if you see lots of New Users and a small number of pages with a short reading time, then it generally indicates either social media or web searches from outside the usual readership reading a single page. Lots of New Sessions with similar page count and time reading tend to indicate more occasional readers popping in. High bounce rates often indicate social media or web searches on specific topics. There are a number of other combinations worth looking for.