Time On Site: a Reasonable Metric? (short take)
By: Jim Sterne, Founder, eMetrics Summit
I assume you don’t have >90 minutes, so – as a public service – here is your TL;DR synopsis of a series of articles Robin Steif brought to my attention with this tweet:
From 6,560 words, down to 1,593. To quote John Hodgeman, you’re welcome.
The leading champions of Engaged Minutes as a worthy metric are Chartbeat, Upworthy and the Financial Times which will soon move from selling impressions to minutes.
“Engaged minutes” is not a panacea but it’s better than unique visitors and page views which put an emphasis on disloyal over loyal readers, encourages publishers to lose focus on their core editorial values and emphasizes “clickability,” or videos of cats.
Social engagement (tweets and retweets, Facebook “likes”) has been a muddle: Difficult to compare and hard to aggregate into a single measure.
The argument in favor of time on site is that a reader’s time is finite. But…
- Assumes longer stories are more valuable
- Is a measure of quantity not quality
- Is hard to tell if a slow reader on a 500 word story is more or less engaged than one who abandons a 5000 word story after 2000 words.
- Has an implicit bias in favor of video
- Encourages reporters and editors to take up more of a reader’s time
The most important business metrics for The New York Times (and similar publications, including the Wall Street Journal and the FT) today are the rate at which subscribers renew.
Positive renewal and conversion decisions are driven, more than anything else, by a handful of stories that strike a particular subscriber as valuable — rather than some amorphous judgment of overall quality.
Measuring Attention Brings Invaluable Insight Into Input/Output Gap
By Daniel Mintz, Upworthy
In response to the above
Upworthy looks at a wide range of signals — mouse movement, keystrokes, video player status, which browser tab is active — to determine whether the user is still actively paying attention to the page.
There are limits to the utility of attention metrics — we agree! We don’t ignore pageviews and uniques and shares, in favor of attention as the only metric that matters.
Measuring attention provides invaluable insight into that gap between input (pageviews and uniques) and the output (shares).
5,000 people loading a page does not equal 5,000 people reading the article.
We found no correlation between how long a piece takes to consume and how much total attention (visitors * avg. attention minutes per visitor) it gets.
We strongly believe that focusing on attention metrics aligns publisher and advertiser in the quest to create the best experience possible for the reader.
If readers don’t like what they see, they bounce. Publishers selling pageviews and uniques, don’t care because they’ve already pocketed their fee. Publishers whose revenue depends on attention, on the other hand, find this disastrous.
The goal is to publish that which creates value for the advertiser, without creating friction for the user.
Engaged time can tell you if the story was read, but not if the politician was subsequently convicted or the law overturned. No one metric can universally quantify something as intangible as impact.
Most digital publishers do not have the luxury of living off subscriber revenues. Only 17 out of 100,000 monthly visitors will convert to subscribers.
Any metric can be gamed but the solid criticisms of engaged time as a currency are not the ones Tofel brings to the table.
A visitor’s default behavior is to leave. 55% of all page views receive less than 15 seconds of engaged time. 30% of visitors will leave without scrolling a single pixel. You bury the lede, they leave. Your headline misleads them, they leave. You pad the story, they leave.
The fight is not between the short form content that receives 30 seconds of attention and the long form that receives 300. The fight is with content that does not receive any attention at all.
The first goal of a brand advertiser is to capture the attention of its target audience. Major studies from Google, Microsoft, Yahoo and Chartbeat show a growing scientific consensus that two primary factors influence success:
– the quality of the creative (who knew?)
– the amount of active time the ad spends in front of someone’s face
An ad that receives 20 seconds of active exposure is more effective at creating recall and recognition than an ad that receives ten seconds of time.
More time = better performance, but with diminishing returns. So yes, the 20 second ad is worth more than the ten second ad, but two ten second ads are worth more than one 20 second ad.
The total amount of engaged time one can capture from a visitor over a given period is far more important than what they do on any single visit.
So, either create great short-form content that encourages people to come back regularly or long-form journalism that maximises engagement from fewer visits.
Business has one huge advantage over editorial: it’s got a built in metric that is standard, easily measured and universally valued. Everyone is happy to talk about dollars.
The attention minutes craze is about catering to advertising and it’s hard to measure ROI on advertising to begin with. It is very difficult to measure the social impact of journalism. Maybe action can help as a metric for impact.
If a story is in a medium like video or interactive that inherently unfolds through time, then measuring time makes sense. But there’s still something to be said for how many people as opposed to how much time. For all its flaws, some variation of page views will probably be important to editorial for a very long time.
Journalism is very much a multi-stakeholder endeavor (sustainability, advertising, impact), so why should we imagine that a single number can capture all aspects of the activity?
Financial Times is well placed with regards to engaged time with loyal subscribers who return often and spend significant time on site.
Financial Times is using engaged time to focus on quality and to provide differentiation between adverts that are not seen, those seen briefly and those that may be seen for a significant duration.
The IAB have introduced a standard for a viewed impression of one second and 50% of the advert in view which has led to adverts that float down the page as users are scrolling.
Financial Times the threshold for an advert to count as viewed at five seconds, setting the bar higher than the IAB standard. A number of campaigns are now running on www.ft.com with brand surveys alongside so as to understand the impact of time in view on changes to brand perception.
Attention minutes have some value, and meet needs that page views or visitors can leave lacking. The question then is, “What is that value of attention and how should publishers apply it to their operations?”
Advertisers don’t care about content or context, as long as they’re seeing results. Attention metrics help solve the challenge of looking at the Web as a direct response medium.
There isn’t one metric to “rule them all” for the audience side. We must create smarter publishing models by including attention among a variety of metrics that indicate audience trends.
Allowing newsrooms, editors and journalists to connect the correct goals of each piece or each publication to the right metrics will help us understand how analytics benefit readers and publishers.
We have found a correlation between returning visitors and the size of a publisher’s site, indicating that these sites may have seen growth and success by doing a better job of understanding their audiences and providing them value. (Full report here: Parse.ly’s Authority Report).
Understanding the audience will always be more nuanced than any quick ROI calculation you can perform.
In a default setup, Google Analytics has two time-based metrics: Average Session Duration, and Average Time on Page. This metric is quite naïve, missing all sorts of important behaviors:
If this is the last page viewed by the user Time on Page for that user isn’t counted.
Time on Page is still ticking for multiple, open tabs open in the browser.
Time on Page is still ticking for a page if the user leaves room.
Google Analytics does have sophisticated capabilities to measure “events”; any type of interaction we can define including:
Only count when the tab is active.
Pause counting if the user pauses scrolling or moving the mouse for more than X seconds
Do not count when the article has scrolled out of view to other content
We still have no way of knowing whether or how much the user is actually paying attention, but using clues like mouse movement and scroll distance do a much better job.
This does require additional tracking code to be installed.
Here are some resources to help:
By: Jim Sterne, Founder, eMetrics Summit
August 21, 2014 - 7:25 pm