Archive for the ‘Research’ Category

In this series, we’ll be looking at what’s capturing attention each month across a wide range of topics – from politics, to science, to entertainment. How does the media cover breaking news and how do audiences engage?  

Bombshell reports this month on Donald Trump Jr., Jared Kushner, and Paul Manafort’s meetings with Russian lawyers dominated media coverage, further illuminating Trump’s alleged connection to Russia in multiple ways. To provide some context around how much of our attention was captured by the coverage on these reports, and see how this month’s revelations compared to previous reporting, we went back and looked at the amount of time readers have spent reading about Trump/Russia-related coverage since February, when Flynn resigned in the first major development of the storyline since inauguration.

It’s no surprise that the major advancements in the Trump/Russia investigation have resulted in large spikes in reader attention, with some even reaching up to 9 times the amount of Total Engagement that initial reports of Flynn’s resignation garnered.

Reader Engagement with the Trump/Russia Narrative since February

As you can see above, the two biggest spikes in reader attention occurred in the same week, sparked by the firing of James Comey as director of the FBI. Engagement on the collection of articles surrounding Comey’s dismissal accrued 203 million minutes of attention during a 5 day span – compared to Flynn’s resignation, which accrued just over 31 million total minutes of attention – with a single day peak of 72 million minutes.

The next story broke in quick succession after Comey’s firing when the President met with the Russian ambassador Sergey Kislyak at the White House and was reported to have both denigrated Comey and revealed classified code-word information. Within a four day period this storyline racked up 224 million minutes of engaged time peaking at 82 million minutes.

To use this as context for last month’s breaking developments, reports of Trump Jr., Kushner and Manafort’s meetings with Russian lawyers alleging to have damaging information on Hillary Clinton accrued 282 million minutes of attention. This story had a much lower single-day peak in attention than Comey’s dismissal or the Trump/Kislyak meetings, however stayed relevant for a whopping 16 days, as opposed to the 4 to 5-day lifespans experienced by the previous stories, highlighted below. This longer story lifespan lead to a higher total engagement for Trump Jr. story.

Reader Engagement with Story Lifespans Highlighted

With all of the revelations and subsequent reader interest during the month of July, it’s no surprise that the number of articles published by media have also been ramping up.

Volume of Media Coverage around Trump/Russia since February

The graph above shows the same Engagement trend line as before, this time overlaid onto the number of articles published about Trump/Russia during that time period. It’s interesting to note that there are a few clear discrepancies between volume of coverage and volume of attention. For example, media coverage saw a huge spike when the obstruction of justice investigation was announced, whereas reader engagement was relatively lower. On the other hand, reader engagement spiked when Sessions’ Russia meetings were revealed, whereas coverage around that time was a little more dispersed.  

Each time a big story breaks, we see attention coalesce toward a small number of narratives. Narratives here are groups of articles that, despite being presented by different publications or authors, for the most part address the same content. For example, after the firing of Comey, all articles about President Trump’s tweets insinuating he had recordings of his and Comey’s meetings would be considered one narrative.

To measure how distributed attention is across narratives we use Shannon Entropy. Points of low entropy represent moments in which user attention concentrates on a small number of narratives, signaling important and evocative developments which unite readers and get them focused on a singular event. Points of high entropy represent instances where user attention is very distributed across all available narratives, indicating a large variety of reader focal points.

As we can see in the graph below, entropy tends to drop with each large spike in attention, clueing us into moments of story development in the Trump/Russia saga. It’s interesting to note that this is particularly the case for the spike corresponding with Congressional testimonies.

Narrative Entropy around Trump/Russia since February

The three points at which attention was the most concentrated (or had the lowest entropy) over the timeline correspond to:

  1. Comey’s first Congressional testimony when it was announced that the FBI was investigating Russian meddling in the 2016 election, and potential collusion between the Trump campaign and Russian officials
  2. Sally Yates’ testimony before the Senate Judiciary Subcommittee concerning Russian interference in the 2016 U.S. election
  3. Jared Kushner’s meeting with the Senate Intelligence Panel on Russia


These dips become especially pronounced likely because of the timeliness of each issue – with each moment, many major news organizations release publicly available transcripts of meetings, such as Comey’s transcripts released the day prior to his testimony before the Senate Intelligence Committee or Kushner’s pre-meeting statements, which allow the public to follow along with each new piece of information. The low entropy we see shows how a singular narrative starts to capture the majority of reader attention.

At a time when political news breaks almost weekly, it’s interesting to take a look at how each narrative compares in terms of coverage and reader interest. How do you see this larger storyline evolving? Will reader interest diminish or will it continue to grow? Let us know your thoughts in the comments below.

The Chartbeat team and I just returned from an incredible trip to Vienna for the GEN Summit 2017 where the three-day action was non-stop. Perhaps most exciting was the opportunity to gather with media leaders from around the world to discuss the biggest issues facing us today.

The main themes of the week focused on how newsrooms around the world have adapted to deal with emerging trends in the industry, including the role of Facebook, the deeper need for data and technology in everyday strategies, the move to subscriptions and exploration of better monetization models, and how we can learn from each other cross-culturally.

As part of the larger discussion around the monetization of news, I participated in a panel titled “How to Monetize Content: Lessons from China.” China is an increasingly interesting news ecosystem to study, primarily because of the burgeoning tech landscape and unique media market there. But is it really as different as we think?

With media heavily regulated by the Chinese government – and with the most prominent platforms in the west, Facebook and Google, blocked entirely – publishers working with Chinese platforms primarily avoid the “news media” label to distance themselves from political news and regulations. Thus, the focus of content providers in China is not a suppressed “Hard News” approach, but instead, a “Soft Content” or lifestyle approach. Social media influencers as well as companies have been able to experiment with content and technology in new ways through this lens to both inform readers and drive revenue streams.

To illustrate this, we look at the top three apps in China, which are all homegrown technology companies. They are:

  1. Toutiao
    A news aggregator that looks and acts similar to the Facebook News Feed, without the social connections. The headlines and content a user sees are automatically selected and ranked based on previous behavior and reading history. Toutiao’s revenue model is primarily ad-based.
  2. Weibo
    Weibo is similar to Twitter with a one-directional following, and the content a user sees is organized based on both users’ selection and actual consumption. Weibo is much more social than Toutiao, and less private than WeChat. The revenue model is primarily ad-based.
  3. WeChat
    Owned by the parent company Tencent which also provides games, e-commerce and digital currency, WeChat offers all content and news separate from the main app experience. Advertising isn’t the major revenue source for WeChat, but content creators can see revenue from micropayments.

Despite the different media environments and platforms in China compared to the U.S. and Europe, the content approach is actually not as different as one might think. While there may be more freedom for publications in the U.S. and Europe to write about politics & society, lifestyle content still dominates traffic in the the West. According to the Chartbeat data below, you can see that the amount of traffic on lifestyle content outweighs that of society and politics. In China, because political content is so heavily regulated, lifestyle content is where the money is really made.


It is noteworthy that in both China and the West, soft content (lifestyle) is relatively easier to monetize. Both native ad formats and e-commerce are more natural extensions of this content type as they can fit more smoothly into the topic at hand.

In summary, we see that the monetization models for publishers that apply across borders are closely tied to the type of content they produce. While hard news is difficult to monetize globally and is highly dependent on the country or political or regulatory environment, soft content may be the place where global lessons – particularly from the platform standpoint – can be learned.

For more on this panel and the many others that took place at GEN, check out MediaShift’s article – Trust and Tech Take Center Stage at 2017 GEN Summit in Vienna.

This article was originally published on DCN. Download the full study here.


Does improving page load time positively impact readership?

The Accelerated Mobile Pages (AMP) Project was launched in February 2016 to address both the increasing bloat of web pages and the subsequent consequences of a diminished user experience for readers, publishers, and advertisers. By creating a platform from which media companies could publish clean, streamlined versions of their articles, AMP promised to speed up the average page load time and make it easier for visitors to stick around and read their content.

But how does it deliver on this promise, and what is the impact on consumer engagement?


The effect on publisher traffic

We pulled actual consumer behavior data across 360 sites using AMP and FIA from June 2016 to May 2017, and our research shows that while usage rates and the subsequent number of articles consumed on each platform differs, it turns out that both AMP and FIA content have been receiving larger and larger shares of publishers’ mobile traffic, and at fairly equal rates. As of mid-May 2017, a typical publisher who implemented AMP saw 16% of all mobile traffic on their AMP content. Comparatively, publishers with FIA saw 14.8% of all mobile traffic on FIA content.

The need for speed

These days, it’s all about speed. In fact, 47% of consumers expect a web page to load in 2 seconds or less and 40% of people abandon a website that takes more than 3 seconds to load, meaning they never reach the published content at all. So how are AMP and IA optimizing page load times?

Chartbeat analysis below shows that AMP loads roughly four times faster than the standard mobile site experience, and Instant Articles load even more quickly – in fact, 88% of Instant Articles load too quickly for us to even register a load time. Now that’s fast.

This is a big deal, and proves that both initiatives are delivering on their key promise: providing a much quicker load time to improve reader experience. And publishers are seeing the effects — with less time spent waiting for pages to load, consumers have more time freed up for engaging with content.


Better user experience, more engaged readers

With so many distributed ways of finding content and such short consumer attention spans, every second counts. So how does AMP stack up?

Chartbeat’s data shows that readers engage with AMP content for 35% longer than standard mobile web content, spending an average of 48 seconds with AMP content vs. 36 seconds with mobile web content when coming from search. The fact that readers are engaging for so much longer than they normally would suggests that user experience really does matter in catching and holding attention.


The future of high speed mobile  

The big question still remains: will publishers continue to scale their efforts and should these high speed platforms become the mobile industry norm?

While the increase in Engaged Time for publishers who have adopted both AMP and FIA are compelling, the jury is still out on whether the end justifies the means. The numbers here conclude that consumers clearly value these optimized mobile experiences and this may just be the evidence we need to validate their potential. However, larger questions still remain around each on both publisher value and quality.

As publishers continue to use these platforms and readers continue to react, it’ll be interesting to see how AMP and Facebook IA evolve in the future – possibly in two completely different directions.

At the end of the day, visitors deserve a fast, clean, enjoyable reading experience where they spend most of their time – on mobile. And publishers can benefit from a more effective mobile environment where they can distribute and monetize content and scale their success. But at what cost? How can we make this win-win?

Download our full study here.

According to recent Chartbeat research, the answer is no.  

The different ways a visitor can land on your content influence not only what they read, but also how they read. Recent research from Chartbeat shows that people exhibit different content consumption behaviors when coming from external platforms like Facebook and Google than when already on your site.  

While exploring the differences in content consumption, we dug into a common myth: do more shares on social media mean more subsequent engagement on the article? What is the correlation between what people share on Facebook and what they actually read?

It turns out, high interactions or virality on social media don’t always translate into actual reader engagement with stories.


To examine the relationship between social engagement and readership, we looked at the 1,000 most shared stories on Facebook for November-December of 2016 and compared their share counts with the total amount of time visitors spent reading them on publisher websites.

While there is a small positive correlation between shares and total engagement, the relationship between the two is quite weak. For stories that attracted 1000 shares on Facebook, the Total Engaged Time they earned ranges from around 14 hours to over 1000 days. This tells us that social media interaction and actual reader engagement are not as closely aligned as many tend to think.

Why is this so important?  It speaks to the importance of headlines on Facebook, and to the need for content creators to evaluate not just social metrics like shares and likes, but also time spent with content, to get a true read on engagement.

For more insights on Facebook and Google behaviors and what they mean for your audience-building strategies, download our latest study here.

Are you seeing similar trends? To talk to us about it in more depth, email us at

This week, we partnered with The New York Times – The Upshot on a study of supply and demand in the news. Essentially, we sought to uncover how many articles news organizations wrote about a given event compared to the demand for these articles among their readers, measured by Engaged Time.

In a time of “alternative facts” and “fake news,” there’s a perception that the media isn’t covering the issues fairly and may be only covering the issues that align with the publication’s political leaning. We wanted to find out if this was really what was happening.

Bowling Green Massacre Article Supply and Demand

We looked at 148 news publishers, divided into liberal and conservative buckets by The Upshot using scores from a study done at the University of Michigan, and studied article supply and reader demand of six recent news topics:

  1. Inauguration crowd size debate
  2. Alleged Bowling Green massacre
  3. Muslim travel ban
  4. Michael Flynn’s resignation
  5. Super Bowl LI
  6. The Grammy Awards

In the end, we found this perception of media bias is not completely true, as publishers on both sides are writing about each topic roughly equally. Instead, this perception may stem from the reader demand side of things; while supply of articles written on each topic was generally consistent across conservative and liberal news outlets, readers on each side do not read the same topics equally.

It’s not media coverage but actually consumer behavior trends that shape what gets read and what doesn’t.

The polarizing perceptions of reality are not only shaped by the content supplies from newsrooms but also by demands from their audiences because, after all, news consumers are the ultimate constructors of their own realities.

Head over to The New York Times’ The Upshot to check out the full study.