Archive for the ‘Data Science’ Category

Scrolling in the Deep

August 10th, 2016 by Katie

To scroll or not to scroll? When 35% of desktop users leave a page before scrolling down at all, you should do whatever you can to capture the attention of the other 65% and make sure they’re actually reading your content.

How readers are moving around on your pages — particularly, how far down the page they’re scrolling — can tell you a lot about how to make your pages more engaging. Take a look at these remarkable statistics about page interactions:

  • The most viewed area of the page is just above the fold. Viewership peaks (just over 80%) at about 550 pixels.
  • The portion of the page below the fold is viewed for nearly three times as long as the top of the page. Pixels at the top of the page are in view for the shortest amount of time — about 4 seconds — and the amount of time in view steadily rises as we move down the page, peaking at around 1200 pixels.
  • Readers who do scroll down past the fold engage much more towards the bottom of the page than they do at the top. We see this represented in the figure below, where we show the amount of time each area of the page was actively viewed by those who scrolled to view it at all.

 

Engagement Across the Page

What does this mean? Don’t just structure your content like every other article out there. Since there’s a large drop-off of readers once you get past the fold, you should touch on your main point early on to speak to all of your readers. When finalizing layout, however, keep in mind how readers typically read and scroll through pages. By setting up your articles to suit these behaviors, you’ll see much more engagement with your content.

Want to learn more about how to improve your layout strategies? Download our checklist for Engaging Article Layout, and get in touch at productoutreach@chartbeat.com.

 

Source: Al Jazeera

Source: Al Jazeera

From the time the UK referendum on EU membership was announced in February, several hundred articles per day were published on the topic of Brexit. This number broke into the thousands on June 13, ten days before the polling, and peaked at over 22,000 articles on June 24 when the Brexit results were announced.

So what can the data around Brexit teach us about how people read the news, what topics capture their attention and how they use news sources vs social and search? For this post, Chartbeat took a look at how Brexit has been covered and read, tapping into data from our network of more than 50,000 publishers, and we uncovered a few interesting patterns:

  • More coverage does not necessarily mean more reader attention
  • The channels by which readers discover stories change during an event’s lifetime
  • Traffic driven by social and search reflects people’s differing interests in Brexit stories


Let’s dive into each.

 

Media Coverage Does Not Always Equal Reader Attention


It’s not surprising that the media gave high priority coverage to this topic , yet public attention wasn’t quickly swayed toward the referendum. After all, 14 months ago the concern over the EU was only the seventh most important issue on voters’ minds according to Ipsos MORI’s survey, noted by
Peter Preston in his Guardian column.

 

The upper panel on the above chart shows how many articles mentioning Brexit were published across the Chartbeat network on a daily basis, while the lower panel shows how many hours people spent reading those articles. The dashed lines mark some important dates related to the Brexit referendum. To many voters, the polling was like a final exam that didn’t gain proper attention until the last minute.

 

hourly_posts_vs_attention.psd

Zooming into media coverage and reader attention, this second chart shows a similar trend on an hourly basis. It illustrates a “pulse” on both the supply and demand side for each day. However, while the media had a strong beat right at the start of the polling on June 23, people’s reading behavior didn’t echo as strongly until much later.

 

Story Discovery Changes During the Lifetime of a News Event


In terms of total attention, June 23 appears to have been another lukewarm day. However, when we break consumption down by referral type, it actually reveals one of the rare moments when
search traffic catches up with social traffic, highlighted in yellow on the following chart. Why is that worth noting? Social traffic is generally driven by passive browsing of news feeds and the like, whereas search traffic is driven by proactive inquiry of specific questions and topics. For that reason, social traffic tends to beat search traffic, as we see on all days other than the polling on June 23 in the following chart.

What’s also notable here is the spiking traffic driven by internal navigation. It indicates that media companies did a great job promoting Brexit stories on their websites and attracting substantial attention from their audience. When the referendum results were announced on June 24, social traffic had a huge jump, which implies people wanted to talk about it for various reasons, such as victorious joys for the Leave camp, surprise and anger for the Remain camp, consequences for overseas jobs, driver’s licences, pensions, and more.

The search spike from the announcement of the referendum results didn’t last: search traffic dwindled on June 24 as information became sufficiently diffused through social media and other communication channels.

 

hourly_referrer_type_logo_annotated

 

Even Regarding the Same Event, Search and Social Readers Consume Fundamentally Different Stories


The last chart further reveals the gap between the supply and demand for particular types of stories about Brexit. The following chart shows how many articles were published in our network around the same story (or group of articles written about the same topic and covered by various news outlets), and how many hours readers spent engaging with each story.

 

*Engagement measured in hours

The stories shown above are the top 20 as ranked by the number of articles published about that story (visualized in the upper panel). Story volume is generally driven by major events, such as the start and end of the referendum, and major forces, such as power struggles among political figures and fluctuations of financial markets.

However, in the lower three panels, we see that traffic volume, search volume, and social volume, measured by engaged hours, differ across stories. Via search, people sought explanations relevant to themselves beyond mere facts. The top searched stories are generally long-form explainers and analyses, such as “what happens if UK votes to leave” and “economic consequence if Leave wins.”

The top social stories have a distinctly different flavor. Not necessarily informative, they carry more emotions, e.g., “regrets and anger about results,” surprises, e.g., “Farage breaks Brexit pledges,” and oddities, e.g., “Brits Google what the EU is.” Unsurprisingly, Mr. Trump buckles up in the front seat on the social wagon.

Brexit news won’t be slowing down any time soon. With Boris Johnson announcing last week that he will not be running for prime minister, the debate around Scotland’s future, and the halo effect of the UK’s decision on the upcoming US election, we’ll have plenty of news to analyze and report on in the coming months.

As many major publishers and platforms have transitioned to using HTTPS, a great move for user privacy and security, a side effect has been a commensurate rise in dark social traffic — traffic that can’t be attributed to a particular referrer. Luckily, sites using HTTPS can still have their outbound traffic properly attributed if they chose to do so (e.g. by use of the meta referrer tag). We’ve chronicled major changes to dark social attribution here to ensure that publishers are up to date on the meaning of their traffic sources.

One of the largest sources of dark social on the web has been the Yahoo homepage, which drives an enormous amount of traffic and moved to using HTTPS over the past year, causing their traffic to become dark for publisher sites that don’t default to HTTPS. For publishers who have partnerships with Yahoo, this has meant that directly attributing the volume of traffic they’re receiving has been difficult.

On June 2, though, Yahoo pushed a change to add a meta referrer tag to their homepage and correctly attribute their traffic to sites using HTTP, and we’ve immediately seen dramatic results, as represented in the figure below.

yahoo

 

Since the change, we’ve seen a roughly 6x increase in attributable traffic coming from Yahoo, making it one of the most significant referrers on the web. On the day before the change, Yahoo was the 16th largest referrer across the Chartbeat network, and in the hours after the change Yahoo jumped to the sixth largest referrer (after Facebook, Google Search, Twitter, Google News, and Bing).

Between this change and other updates by LinkedIn and Facebook, we’ve seen significant moves in the past 18 months by many of the world’s largest platforms to ensure that all traffic is correctly attributed. We’ll continue to work with publishers and platforms to track down sources of dark social, and we’ll keep you updated here as more publishers move into the light.

Technical note: for those used to the hsrd.yahoo.com referrer, traffic sent via this change will carry the referrer yahoo.com, not hsrd.yahoo.com.

Two weeks ago, Facebook opened up its Instant Articles platform to all publishers. And last week, Facebook announced that they will be updating their News Feed algorithm once again. The most recent change to their algorithm will look at predictions of whether a user in the Facebook mobile browser or on an Instant Article page will click into an article and actually read that article. Time spent viewing the article will continue to be a large factor in News Feed rankings.

When Facebook makes changes, the publishing industry reacts with questions and concerns (see, for example, here, here, or here). That said, each time we here at Chartbeat have looked at Facebook referrer traffic in response to one of these changes, we haven’t seen any major effect across our network.  Here’s what the median percentage of traffic from Facebook looks like across our network so far this year:

graph_04_25_01

Besides the typical weekday / weekend variations, traffic from Facebook is remarkably stable. We see Facebook driving between 40-50% of traffic on mobile devices during peak weekday traffic and about 12-15% of traffic on desktop devices during peak weekday traffic (note that these numbers exclude cases in which we have no data for referrer, as is the case for dark social). Even given the two big changes that happened this month, we are actually seeing a slightly higher-than-normal amount of referral traffic! This increased traffic is on the order of 3-4% for the median (smaller if you look at the average; 1%).

It is important to note that these curves show the median proportion across our network. Individual sites respond in different ways, so you may have seen your traffic rise or fall in response to one of these changes. Sitting in the newsroom, it is hard to see the forest for the trees, so to speak — we have the power of statistics on our side. But from what we continue to see, the majority of publishers are incredibly adept at responding to Facebook’s changes and are keeping referral numbers high.

Beginning several days ago (the evening of Tuesday, 1/20, to be precise), you may have noticed a significant increase in the traffic on your site from LinkedIn: across our network, traffic from linkedin.com increased by over 3x. Below, we’ll detail why that change occurred, and what publishers should expect going forward.

Over the past year, publishers have become increasingly interested in traffic from LinkedIn, as the LinkedIn team has been steadily working to improve their feed experience with the launch of their new mobile app and content platforms. Nevertheless, when looking at referrer traffic in analytics tools like Chartbeat, web traffic from linkedin.com has always seemed smaller than it should for such a large platform, especially given the volume of traffic we see from LinkedIn’s counterpart apps, which shows up under the referrer name lnkd.in.

On January 20th, that changed when LinkedIn made a change to correctly attribute their traffic, some of which had previously been categorized as dark social. The impact of that change was immediate and significant.

Let’s look at traffic coming from linkedin.com to sites across the Chartbeat network over the last six months, we see two trends: a steady increase over the year, followed by a huge increase at the end of January.

linkedin_01_v2
Zooming in on the right side of the graph, January, 2016, we can see the immediate change in traffic as the attribution change was pushed:

linkedin_02_v3If we compare numbers from just after the change to the same time during previous weeks, traffic from linkedin.com was up by over 3x.

Some sites saw more than 6x increases in their LinkedIn traffic.

While LinkedIn still isn’t a major traffic source for many types of sites, we expect that many business-, media-, and technology-focused sites will see LinkedIn as a top-10 referrer going forward.

With Facebook’s change last year to help attribute all of their traffic, LinkedIn’s change here, and other work to come, we’re excited to see more traffic correctly attributed. We’ll continue to work with platforms in the coming months to bring their dark social traffic into the light.