The Horror that is Clickbait

Posted on

A friend linked me to an article by Ben Collins on the Daily Beast this morning. It’s not a website that I usually visit, but the headline had me interested: “No, Spooning Isn’t Sexist. The Internet Is Just Broken.“.

I’d been seeing references to the clickbait article that Collins was referencing in his headline crop up all over the place and thought how ridiculous it was but I hadn’t actually read it. The Daily Beast article made sure that I didn’t have to.

The important thing about Collins’ piece was that we come up with a new measurement to replace the “unique visitors” metric that the Internet seems so keen on. Doing so will mean that, in theory, clickbait will no longer be a thing because success will be measured on engagement rather than clicks. I’m pretty sure no-one actually reads the whole of these “articles”.

Collins pointed out that this new measurement should reflect time spent on the site. This is good because it helps to measure actual engagement with the content (in the case of clickbait, if an article is being read or it was simply clicked on due to the outrageous headline), but it’s also difficult to work out.

Tony Haile told me last year it’s, in part, a measurement called “Time on Site.” He’s been cooking up a formula for it ever since, trying to measure how long someone spends on an article.

This got me thinking about my interview with SessionCam a few months ago. I’ve briefly mentioned the technology they’re working with previously, but I’ll state it again (because, despite not getting the job, my inner nerd thinks it’s super cool). They record the session of a user in terms of what links they click on, where they move their mouse pointer etc., and then play what they’ve recorded back over the top of the website so it looks like a video. Now, if they’re  recording this, they must be able to see the timings too, otherwise it wouldn’t be a real-time playback – you’d just see the mouse moving from one place to another seemingly instantly.

I don’t know if they do this already, or plan on doing so, but I’m thinking that this could be used to help with this new measurement of user engagement. You wouldn’t be able to tell if someone was reading an article per say, but you could probably assume they are based on if they are scrolling slowly down the page, or spending a reasonable amount of time on that page. With the scroll measurement, you would also be able to take care of the problems that Collins pointed out in his article. It should be relatively easy to spot people who have clicked on the article but haven’t read it yet, or those who have just loaded the page and left their computer.

As the boyfriend works for SessionCam, I asked if it was possible with the technology. He said it was (and he also said it was a good idea – even though he’s totally biased, I’m pretty sure he wouldn’t say it was a good idea if it wasn’t) so I’m pretty pleased with myself there. Sometimes, it’s almost like I understand all this technical stuff. Sometimes.

I’m sure there are other ways of working out actual engagement rather than relying on click rates, it’s just finding something that’s easy to do and can be implemented on a large scale. Whether or not that involves something like what the people at SessionCam are already doing is another matter.