Twitter  "Algorithms have consequences." Zeynep Tufekci on Ferguson and net neutrality: nie.mn/VpE1Ef  
Nieman Journalism Lab
Pushing to the future of journalism — A project of the Nieman Foundation at Harvard

More data on The Daily: What’s prime time for iPad use? And which stories get tweeted the most?

Yesterday, I wrote about a fascinating set of data I’d obtained with the help of the folks at PostRank: Every tweet generated from within The Daily iPad app from launch through March 31.

I used that data in yesterday’s piece to try to learn about the relative size of The Daily’s audience. And it showed that the number of tweets generated by the app’s users has been declining ever since launch — which would seem like an indicator that The Daily’s audience might not be growing as fast as some might like.

But the data’s much richer than that, and I’ve been spending the past few days digging into it to see what else might be of interest from a future-of-news perspective. So today I present three additional findings that try to answer these questions:

— Do people use iPad apps at different times of the day than they use other, non-tablet devices?

— What kinds of stories in The Daily get shared the most — or the least?

— Does the distribution of tweets among Daily stories or among Daily readers match a power-law distribution?

More evidence that tablets are used at different times than the general web

This might be my favorite chart from the data set. The blue line you see above shows the timing of tweets generated within The Daily’s iPad app, tied to hours of the day. So the x-axis starts, on the left, at midnight, the middle is noon, and the right side goes up to midnight the next day. (The time zone here is set to Eastern time, which is where The Daily’s based and where the largest chunk of its users are likely to be.)

The green line, in contrast, shows the timing of all tweets to thedaily.com that aren’t generated from within The Daily’s iPad app. So that would include anyone using the Twitter web interface, a mobile Twitter app, or a desktop app like TweetDeck. (Both lines show the share of each group’s tweets happening at that hour — not the raw totals.)

This actually approximates a broader device divide. All of the tweets represented in the blue line came from iPad use; the vast majority of tweets in the green line came from desktops, laptops, or smartphones. And notice how interesting the patterns are! The tweets from the iPad app peak in the morning (around the 7:00 and 8:00 hours) and again in the evening (around 8:00, 9:00, and 10:00). The non-iPad app tweets follow a very different pattern: They match up with the work day, highest from 9-to-5 and peaking just before and after lunch — in other words, when lots of people are sitting at a desk looking at a computer.

Within the app, The Daily’s pattern looks an awful lot like the old newspaper consumption pattern: in the morning before work and in the evenings after it. (There used to be this thing called an afternoon newspaper, see.) It also looks a lot like what Read It Later found in the usage of its iPad app, which was similarly bimodal. (Although their data was more slanted toward the evening — it makes sense that The Daily, which is published early each morning, would be relatively stronger in the a.m.)

The non-app Twitter traffic, meanwhile, matches up well with the work that Pablo Boczkowski has done (in News at Work and elsewhere) on how the shift to consuming journalism in an office environment might be impacting the field.

I can’t say how, exactly, but it’s increasingly clear that a successful tablet news app will need to somehow be optimized for a leisurely, at-home, lean-back environment.

News stories get shared more than other sections’

The Daily does a good job of making the section each story is in clear from its URL. In other words, it’s clear that

http://www.thedaily.com/page/2010/12/12/020411-arts-beavis-butthead/

http://www.thedaily.com/page/2011/01/01/031111-news-tsunami-video/

and

http://www.thedaily.com/page/2011/02/01/020211-sports-wolken-1/

are in the Arts & Life, News, and Sports sections, respectively. With that in mind, I wanted to find out what types of stories were most likely to be shared via Twitter — the flashy Gossip section, the image-heavy Sports section, or something else entirely?

I determined the section for all 6,024 tweets generated from within The Daily’s app that linked to a story page. Most I could determine quickly from the URL; a few hundred I had to determine manually for myself. Here are the results for each of The Daily’s seven sections:

News: 3,427 tweets (56.9 percent of all in-app tweets)

Business: 33 tweets (0.5 percent of all in-app tweets)

Gossip: 289 tweets (4.8 percent of all in-app tweets)

Opinion: 719 tweets (11.9 percent of all in-app tweets)

Arts & Life: 559 tweets (9.3 percent of all in-app tweets)

Apps & Games: 630 tweets (10.5 percent of all in-app tweets)

Sports: 367 tweets (6.1 percent of all in-app tweets)

So The Daily’s News section accounts for an absolute majority of the tweets sent from the app, while the Business section is almost invisible to Twitter.

Of course, there’s not an equivalent amount of material to tweet in each section — News is a lot bigger than Opinion, for instance. So for the past three days, I’ve been counting by hand the number of story pages in each section. (I would have liked to have a larger sample, but The Daily doesn’t do archives. And note this is horizontal-swiping story pages, not stories. Many stories have more than one page; many pages have more than one story. The story page is what the tweet button in the app uses to generate the link URL.)

Here’s my count of the total number of pages by section since Monday:

News: 101 pages (32.3 percent of story pages)

Business: 17 pages (5.4 percent of story pages)

Gossip: 24 pages (7.7 percent of story pages)

Opinion: 27 pages (8.6 percent of story pages)

Arts & Life: 28 pages (8.9 percent of story pages)

Apps & Games: 31 pages (10.0 percent of story pages)

Sports: 85 pages (27.2 percent of story pages)

Interesting! So — assuming that this section breakdown has been largely constant through The Daily’s two-month run — the News section is generating 57 percent of the tweets with only 32 percent of the content. Meanwhile, Sports takes up 27 percent of the content but generates only 6 percent of the tweets. I don’t know that there are any larger conclusions to be drawn here, but the wide variation sure is interesting.

A classic pair of long tails

Anyone who’s hung around the Internet has heard of the long tail, a.k.a. the Pareto principle or a power law. (Those aren’t strictly interchangeable, but they’re often tossed around as if they were.) For example, in the music business, a few blockbuster albums and songs have usually made a huge portion of the industry’s profits, while a huge number of other albums and songs barely sell at all — the long tail.

When I started crunching the numbers, I saw two nearly perfect cases of this kind of distribution. First, I looked at which Daily articles were tweeted the most within the app. A total of 2,139 unique URLs were tweeted at least once. I ranked them from most- to least-tweeted and this chart is the result:

That’s basically a perfect power-law distribution. A few of the most-tweeted stories on the left get (relatively) huge amounts of attention, while nearly half of all tweeted URLs — the long tail — were tweeted only once. (And, of course, many were never tweeted at all.)

In case you’re wondering, here are the 10 most-tweeted stories by Daily app users:

1. The introductory “welcome to The Daily” editorial from Feb. 2, launch day: 52 tweets

2. A video story of a winter storm, also from Feb. 2: 37 tweets

3. A nine-sentence story on a Catholic confession app from Feb. 8: 36 tweets

4. A six-way tie between a review of Portlandia (Feb. 2), an Egypt infographic (Feb. 2), a two-graf gossip piece on an NBC firing (Feb. 5), a piece on new Army social media guidelines (Feb. 6), a tsunami video (March 14), and a Japan relief appeal (March 16): 30 tweets each.

10. “American military commanders have declared war on pets,” Feb. 13: 29 tweets

You’ll note that four of the top 10 are from launch day, three others are from the first week, and only two are from the past month.

The other long tail to pop out from the data came when I wanted to see how many tweets each Daily user was generating. I gathered up anyone who tweeted at least once from the app, then sorted them by how many times they’d tweeted — most on the left, least on the right. Here are the results:

As you can see, a very similar result: A few users tweet a lot, but most — just over 70 percent in this sample — tweeted only once from the app.

The most frequent Daily twitterer is @amadorn (“We buy and sell vintage and contemporary costume jewelry”) with 71 tweets from The Daily app over the two-month period. Nine other Twitter users tweeted more than 20; another 34 tweeted more than 10. In total, 3,316 people tweeted from the app.

I’m not done mining this data quite yet — there are still some interesting nuggets to share. Like: What can we learn about the people who use The Daily from their Twitter bios? Where are they tweeting from? How does in-app tweeting differ from out-of-app tweeting? More to come.

                                   
What to read next
BuzzFeed HQ
Caroline O'Donovan    Aug. 11, 2014
With a new round of investment from Andreessen Horowitz, the viral giant aims to get even bigger. One step down that path: making it more clear to readers what kinds of content BuzzFeed really offers.
  • Chuck

    I still haven’t figured out why Gossip seems to be such a big part of the plan at The Daily. And these numbers don’t seem to show that there’s a lot of interest (at least among the Twitter users out there).

  • http://twitter.com/JoshDehaas Josh Dehaas

    Interesting.

    This is what I see. The stuff that does well for the daily on the I-pad appears to be the same stuff that does well on the web: surprising videos, infographics and quirky “would you look at this” bits.

    That should be a lesson for any magazines that try to just give us their print edition in iPad form.

  • Anonymous

    The EIC of The Daily is formerly of the NY Post. The Publisher came to The Daily from MTV’s digital group. I think that could explain the Gossip angle. Also, I’m sure they’ve seen how well gossip has helped The Huffington Post lately.

    Personally, I think they should eschew the Editorials (maybe they have since I stopped paying attention when the app kept crashing on me in the first two weeks).

  • Paul Colyer

    First, I think what this data shows, more than anything else, is the wide readership of the daily as opposed say, to daily newspapers. For newspapers and magazines to survive they are going to have to tailor their content to their audience, very similar to what advertisers are now doing on the Internet. Likely two versons of printable formats available; the usual newspaper formats that we see now to be read on laptops, desktops, IPads and other similar tablets. Shorter, headlined featured topics available on mobile devices such as IPhones, IPads and Tablets. Content for mobile devices will be and is similar to what we see on Internet search engines, such as Yahoo, MSN and AOL.

    Second, It isn’t as much about the difference between IPad and laptop users as it is about the easy of use of IPads, tablets and mobile phones versus laptop and desktop computers. As IPads and other mobile devices become more acceptable within the office, readership trends will equal what The Daily has currently discovered. That is, readers within office settings will still uses IPads, tablets, etc. during the same core hours that as demonstrated in the current study.

    In summary, what has clearly changed is the ease of use of mobile devices versus laptops, desktops and even the print media, i.e. newspapers. We are an increasingly mobile world and workplace with mobile devices reflecting this trend. There will be multiple platforms for media all media outlets but especially for magazines and newspapers. This trend has already been set via e-books, e-newspapers and e-magazines. TV is adapting as I write this. Radio and radio content, while much slower to adapt, will play catchup.

  • http://twitter.com/nikkiusher nikkiusher

    This is an obscene crazy effort. Props.

    Question here that’s important to the results: what counts as news? because it could be that the daily’s definition of news is different from what say, the wall street journal might define as news. Would the Royal Wedding be news or gossip? Just something to keep in mind before getting too excited about the tweets that represent news. It could be that news isn’t really news but the good gossip that makes its way to news.

  • http://lavrusik.com Vadim Lavrusik

    I hate to say this, but tweets aren’t everything. The point about most tweets being generated about news content isn’t surprising. Think of yourself as a user. You may be reading gossip, but you’re not going to share because of the perception. So the question is, what is actually happening on the app? Twitter can only tell us so much, and Facebook shares provide a whole other perspectives in terms of what type of content is shared most.

  • Roberto

    What’s wrong with The Daily? Paywall.
    They should take a look at Brasil 247, the first iPad newspaper in Brasil, which is free

  • Myles

    Tear Down That Wall Mr. Murdoch. Free content supported by Advertising & Sponsorships is the way to go. Network TV had it figured out 60 years ago. Why are we trying to reinvent the wheel? Is it that too many “new-bees” are running new media these days? Bring in some seasoned pros. MMF

  • Pingback: » More data on The Daily – What’s prime time for iPad use? And which stories get tweeted the most ? » Nieman Journalism Lab » Pushing to the Future of Journalism Media Strategery

  • http://www.niemanlab.org/ Joshua Benton

    Thanks, Nikki! You raise a good point: I think a better name for the section would be something like News & Features, since it includes a lot of softer stuff (along with, to be fair, a bigger-than-I-expected chunk of international news). The royal wedding would almost certainly qualify as News rather than Gossip in The Daily. But I don’t think that broader definition of news is necessarily that different from that of many other news organizations — the royal wedding’s going to be on page 1 everywhere.

  • http://www.niemanlab.org/ Joshua Benton

    You’re right, Vadim, that tweets aren’t everything, and that The Daily’s data isn’t particularly generalizable. But I’d note that the fact that Business did so poorly would cut against the idea of people disproportionately sharing high-status stories.

    Another hypothesis would be to look at how sharing is impacted by the position of the story in the app. Unlike most news apps, The Daily favors a newspaper-style flow of starting at the beginning and moving through the paper in a single direction. News is first and Sports is last, which might be part of why News gets shared more. (But Business is second, after news, so that cuts against the idea.) I think there’s plenty of room for other analyses, and I hope people do them.

  • Pingback: Hábitos de uso del iPad, según datos de la aplicación TheDaily : Noticias de tecnología – Tecnodatum

  • Pingback: 雜誌電子化再傳失利?iPad付費日報The Daily銷售不盡理想 - Inside

  • Pingback: Tablet Strategies for Publishers: Framework for Content and Form « claus enevoldsen

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | big news

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | CashKlick

  • Pingback: Does Social Media Encourage Quality Journalism? « Chris Hoose

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | AppsPlanner.com Blog

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | Circle Social Media Blog

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | Technology Story

  • Pingback: Los social media propician una mejor calidad de periodismo « Punto de Vista @SarkASStiKo

  • Pingback: Social Media and Quality Journalism « The Social Media Days

  • Pingback: Why Social Media Reinvigorates the Market for Quality Journalism | Business Small Business

  • http://twitter.com/designonmymind Pradeep Nayar

    Interesting…..

  • Pingback: Social Media Reinvigorates the Market for Quality Journalism — ByrnesMedia

  • Kathy

    The “serendipity” and “discovery” of news, as Murdoch put it, is what holds a society together. As we continue to refine what we read/view – because we can – the universe narrows.
    As a pure guess, I would think it highly unlikely that someone would twitter a Daily social (arts, gossip, opinion) link that was simply different (serendipity, discovery) vs in agreement with or totally at odds (outrage) with the personal/political views of the person initiating the tweet.
    The Daily’s introduction video promo focuses on Arts, Sports, pretty images and puzzles – yet from your analysis, news is what we want (which is at odds with universe narrowing…).
    The question left unanswered, so far, is who are “we” at the Daily? Does a jazzy app on a new platform pull in new sustainable demographics that newspapers (and therefore their advertisers) desperately need?
    Thanks for making the effort parsing the data.

  • http://twitter.com/stephen_tracy Stephen Tracy

    Great work Joshua. There are definitely short comings with the data set, but you’ve made the disclaimers, and have done an excellent job demonstrating what can be done with publicly available data. I loved the blue and green chart that shows the twitter usage by the publishing source and time of day. You mention that all tweets are aggregated to EST. Does PostRank provide any geo-location features. It might be interesting to add a 3rd variable to this table, which factors in the poster’s  actual time zone. This way you could isolate the tweets by their respective time zone, and replicate this test for each and compare (assuming you have a large enough sample size). Regardless, both this and the previous post on The Daily tweet stats are fascinating. Great work!

    I know this is a really late comment, better late than never I guess:)