Time spent on a site or a visit ranks right up there with total page views and monthly unique visitors as widely quoted metrics masking as indicators of success for news organizations.
No, it's not a crime to misuse a metric, but isn't a shame to waste your time on something that's not absolutely essential to your site's success?
Plus, the way that time spent is calculated is flawed. All web metrics are flawed somewhat, but time spent is really misleading.
More on the ugly methodology later – let's tackle time spent's uselessness first. In other words, if the methodology were acceptable would time spent still be a key performance indicator?
Advertisers have always made decisions based on the level of engagement a news org's audiences have with its brand and content. But both content and the ways people use and interact with content are different – and thus the way engagement is measured is different, too.
In the past, time spent was an important measure of engagement for news orgs and advertisers. People spent whole chunks of time with one medium or another. Readership surveys measured time spent per day or per week.
Because these were surveys, time spent was based on self-reported information. It was what people said they did vs. what they actually did.
But it didn't matter whether what people said matched with what they did. What mattered was how engaged people felt. People who reported they spent an hour a day with Monday's newspaper but actually only spent twenty minutes believed they spent a large chunk of time and attention with a news org.
In stark contrast, web advertising decisions depend on knowing actual behavior as reported via rows upon rows of numbers ruthlessly pouring out every second. Among many other things, advertisers track the number of times their ads come up and are clicked upon. Sites and audiences are more niche and are highly segmented. The algorithms for and definitions of "engagement" vary for every site and every company.
Time spent just isn't a good indicator of engagement. Someone who spends five minutes a day on a site, goes to five different stories each visit and adds comments twice a week is clearly more engaged than someone who comes onto a site for 30 minutes a week and clicks idly on a few pages while talking on the phone.
How many times have you spent 30 minutes or so on a site, flipping and flapping through what seems like a million page views in a fruitless attempt to find something? Maybe you spent 30 minutes in such a visit once – and never went back.
A news org's success in the long-term will be based not on how much time people spend on a site but what they do once they're there.
How Time Spent on a Site is Calculated
The time spent during a visit is the amount of time between a visit's first page view and its last page view.
- Don goes to the home page of a news site at 8:00 a.m.
- He looks at the stories on the home page until 8:05 a.m. Then he clicks on a link to a local news story.
- Don glances at the local news story page but then gets up and goes to the kitchen to feed his cat.
- He comes back to his computer at 8:10 a.m., and skims the local news story.
- At 8:12 a.m. he clicks on "Sports" in the top navigation bar. He scrolls down the Sports section front, reading almost every headlines and brief.
- At 8:15 a.m. he remembers he has to pay his mortage online. He goes to his bookmarks list and clicks to http://www.thebank.com.
Time on site, as calculated by Omniture or Google Analytics:
Actual time on site: We don't know! It's at least 12 minutes, but it may be longer.
This is what happens, technically.
1. The news site's computer time-stamps Don's visit with a start time of 8 a.m.
2. The computer time-stamps the page of the local news story at 8:05 a.m.
3. The computer time-stamps the page of the Sports section front at 8:12 a.m.
Time spent is the time in between the first and the last page views. The first page view is the home page, with its time stamp of 8 a.m. The Sports section page is the last page and has a time stamp of 8:12 a.m.
Don's bank knows what time Don came to http://www.thebank.com, but the news site doesn't. The news site knows that the Sports section front was the last page, but it doesn't have access to http://www.thebank.com's web metrics data.
For all the news site knows, Don could have spent three hours on that Sports section front. Or maybe he spent two seconds on the page and immediately bounced off because he'd rather pay his mortgage than read one more biased rant from the site's football columnist.
Summary of time spent methodology problems
- You don't know how much time is spent on the last page.
- You don't know if the visitor was actually reading the page. The time Don spent skimming the local news story was anywhere between two seconds and seven minutes. (You really don't know whether Don got up and fed his cat.)
- Bounces – or visits where the visitor only went to one page - are counted as zero seconds.
Most news sites have home pages packed with stories, headlines and briefs. If Don had spent 15 minutes just on the home page before going to http://www.thebank.com, the visit would have counted as zero seconds. The computer needs time stamps on at least two page views to calculate time spent on site.
Time spent? FAHGEDDABOUTIT.