If you’re going to use data to assess how your content is performing, you need that data to be useful, understandable and actionable. After all, it’s no good compiling a report of impressive-looking numbers if you’re not quite sure what they actually mean. Here’s a great question we received recently from one of our users:
‘Why are the numbers of page views (from Google Analytics) and article reads (from CI) different? Aren’t they the same? And, if not, why not?’
If you’re comparing performance data from Google Analytics with our tool, chances are the numbers under GA’s Page Views and CPI’s Article Reads don’t quite line up. Article reads will be lower. This doesn’t mean that one set of data is wrong, rather it points to something else – namely that the underlying way of measuring page views and article reads are different.
A page view is a browser event. This has nothing to do with window shopping – though it can be just as fleeting. Page views occur whenever the code for a page is loaded, so even if an article is loaded in the background (or, the ‘open in new tab’ option), triggered by a bot and even if it lasts for only a few seconds, it counts.
We have our own definition of an impression, which we call an ‘article read’. Now, we’re the first to acknowledge that isn’t the best name as it implies something else, but it’s a legacy name, so we’ve kept it.
One article read = somebody opened a page, spent at least 10 seconds on it, the page was in focus and there was an actual human behind the screen.
How do we know there’s a human behind the screen? Good question. We look for activity on the page: things like mouse moves, scrolls, clicks, selecting something – evidence that there’s some kind of interaction. We think looking at impressions this way gives you a clearer idea of genuine interactions – is there really any value in a counting a page view if it’s so fleeting as to be practically unseen? We’re not convinced there is.
So, while we’re here, let’s also look at the difference between GA and CI when it comes to attention time…
Google Analytics measures ‘session duration’ (which was formerly known as ‘time on page’), which sounds like attention time, but isn’t. What it is is the time clocked between the first page opened and the last page opened. Confused? Well, you’re not alone.
If you were to open Page One at 9am, Page Two at 9.05am and Page Three at 9.15am, the GA-calculated session duration would be 15 minutes.
Obviously there are problems with this:
- If the visitor bounced (left the site after the first article), the time isn’t counted at all
- Time spent on the last session (page three in the above example) isn’t counted
- Any attention time that didn’t result in opening another article isn’t counted
- If you inadvertently leave an article opened in another tab in the background for an hour, this will count, as long as at least one article was opened after
- So, GA doesn’t care about user activity on site
How does Content Insights measure attention time, then?
We look at ‘true attention time’. Just as the name suggests, this is the total time spent attentively by visitors on an article (already there’s a significant difference, right?).
We only calculate the time when the content is in focus and there’s sufficient activity indicating the reader is conscious and reading, not asleep at the monitor or off making a cup of coffee.
But what happens to true attention time when the reader is seized by the urge to re-caffeinate mid-article? Well, if an article is opened in the tab in the background for 20 minutes, then put into focus for seven minutes, followed by a period without activity for five minutes while you go and make that cup of coffee, and then a further eleven minutes of active reading (it’s a really great article), then Content Insights would calculate the attention time as 18 (7 +11).
In Google Analytics this same session would either read as 0, if it was the only or the last page in the session, or 43 minutes, if there was a page view after it.
Doing things this way we’re able to see how readers engage on an article-by-article basis, rather than by taking a view of a session. If you’re trying to work out how articles are performing, this is a subtle difference worth understanding. Of course, you’re likely to use different analytics tools for different purposes, but it’s important to understand which does what so you can use the resultant information in the most useful way possible. Don’t assume that because measures sound the same from one application to another that they are – often they’re not.
We hope that’s cleared that up. Have another question for us? Ping them in this direction and we’ll happily answer them for you. Knowledge is power, right?