> The per-page time is actually quite high; often over 5 minutes on average for a lot of my stories even when its a big crowd. I suspect some subtle misreading of Google Analytics on my part, or else people might actually be reading the whole thing carefully!?
That might be because a lot of people do what I do: keep the article open while browsing comments on HN or Reddit, and switch back and forth between the two tabs while writing my own comment. In fact, I'm doing it right now. I don't know how Google Analytics would measure time spent on a page when the page sits idle in a background tab for the most part but brought to the foreground from time to time.
[Edit] Things get even more complicated if you open 5 articles and 5 comment pages at the same time, and take 30+ minutes to go through all of them.
I tend to have one window for recreation browsing and at the start of the day I will open HN and run down the front page middle clicking the link and the comments pages for anything that looks interesting or highly up-voted.
This gives me a queue of stuff I can read in free time throughout the day but I imagine it really does skew the analytics.
This is why I wrote my own datapoint gathering script (I started it while at Techcrunch, curious about who scrolls down to comments, etc.).
It measures body focus, number of times unfocused, total focus time, scroll position, scroll rate and a bunch of other things.
It isn't completely working cross-browser yet, but it is close and useful enough where for Safari and Chrome visitors I can see how far down the page they are scrolling and how long the page is in focus for.
No doubt, this is part of the reason this noscript_using_reader stopped visiting TC. Call me a tinfoilhat-guy, but I simply would not consent to the collection of such data. It's ironic that I've learned so much from your analysis of data collection techniques elsewhere.
I have been meaning to clean it up and post it to my GitHub, I will do it this weekend. If you like, email me (in profile) and i'll ping you as soon as it is up
A news site in Australia here seems to know when the tab is visible and auto plays the video content for a story. I assume it's using these Javascript events too.
But that can still lead to inaccurate measurements of "time on opage" if the user switches back and forth between tabs. Do you only measure the periods during which the tab is active (and if so, how granular?) or do you simply subtract the first active timestamp from the last active timestamp?
That might be because a lot of people do what I do: keep the article open while browsing comments on HN or Reddit, and switch back and forth between the two tabs while writing my own comment. In fact, I'm doing it right now. I don't know how Google Analytics would measure time spent on a page when the page sits idle in a background tab for the most part but brought to the foreground from time to time.
[Edit] Things get even more complicated if you open 5 articles and 5 comment pages at the same time, and take 30+ minutes to go through all of them.