Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 164

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 167

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 170

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 173

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 176

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 178

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 180

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 202

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 206

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 224

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 225

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 227

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 321

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 321

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 321

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/includes/class.layout.php on line 321

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/admin/class.options.metapanel.php on line 56

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/admin/class.options.metapanel.php on line 56

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/admin/class.options.metapanel.php on line 56

Warning: Creating default object from empty value in /home3/freesms/public_html/LLsocial.com/wp-content/themes/platform/admin/class.options.metapanel.php on line 49
Alexis Madrigal’s “Dark Social” premise is flawed. | LL Social

Alexis Madrigal published a piece to day on how people find content on the web. While I agree with much of his piece including the contention that the web was social long before social media media (think IM, forums, email, etc), and that a large portion of visitors to sites cannot be tracked to the source, I believe he was led the wrong way by his “expert” Josh Schwartz of Chartbeat.

While Alexis acknowledges in this piece that mobile  apps could account for part of this unatributed traffic, he provides a footnote with comments from Schwartz:

Chartbeat datawiz Josh Schwartz said it was unlikely that the mobile referral data was throwing off our numbers here. “Only about four percent of total traffic is on mobile at all, so, at least as a percentage of total referrals, app referrals must be a tiny percentage,” Schwartz wrote to me in an email. “To put some more context there, only 0.3 percent of total traffic has the Facebook mobile site as a referrer and less than 0.1 percent has the Facebook mobile app.”

How is mobile defined here? If they are just looking at screen size I guess is it possible only 4% of The Atlantic’s viewers are on phones, although that seems very low (web in general 16%, news sites 7% reported nine months ago). There is no mention of tablets, iPad specifically. Which leads me too…

What is missing from the analysis is Mozilla Web Kit.

What type of apps could be sending content to The Atlantic? It isn’t much of a stretch to think that the referrer might be one of dozens of popular news reader/personalization apps (think Zite, Flipboard, etc). Those apps generally use their own browser to display pages, and those browsers are almost 100% using a web kit.

As I detailed in my article on underreporting of Pinterest traffic, the majority of apps use Mozilla Web Kit to provide a more integrated experience, and none of these app browser provide referring data. Once you include iPad in your numbers, I have to believe a significant portion of this untracked traffic is coming from apps, and it doesn’t just have to be news apps.

Social traffic is underreported, but much of it comes from the big social networks.

To quote my previous piece:

Jim Gianoglio, Manager of Insight: Social & Mobile at LunaMetrics, told me that many app visits will show up this way. He indicated that Facebook has figured out a way to resolve this issue with their app, but that Twitter, while better tracking referrals with the t.co link shortener, still sends traffic from their own apps  (as well as many third party apps) without clear referral attribution.

My own vist to Alexis’ piece would have fallen under this “Dark Social” traffic. I was using Hootsuite on my mobile phone when I saw a tweet with a link to the article. I clicked on it and read it using Hootsuite’s browser that use web kit. So this “Dark Social” traffic includes, at least in my case, traffic from mainstream social media.

The purpose of this piece was to clarify that there isn’t an unknown majority of people who forgo social networks to share links. Certainly some people email links or share them in chat, but to attribute around 50% of The Atlantic traffic to this group of people isn’t even close to the most likely scenario.

 

 

2 Responses to Alexis Madrigal’s “Dark Social” premise is flawed.

  1. avatar Josh Schwartz says:

    I think that you’re completely right that there’s a long way to go toward being able to fully analyze mobile traffic. Alexis and you both note that traffic from mobile apps can be counted as “dark social” because it can lack a referrer, and there’s no question that that’s correct. But the issue of how mobile traffic should be measured is, in my mind, a different discussion.

    The issue of mobile came up because Alexis wondered if the issue you identified — apps referrals reporting as dark social — was a possible cause of the large amount of dark social traffic that we measure. And, the point of my comments was to provide context on our numbers: if our numbers simply aren’t drawn from mobile traffic, then no matter what the referrer is recorded as for mobile users, it won’t throw off our referral numbers in a significant way. And, the fact is that for the sample I measured for Alexis, only 4% of traffic was from mobile (where “mobile” is defined based on OS). You raise a valid point — if traditional metrics say that mobile accounts for more than 4%, why was only 4% of our data drawn from mobile? The answer lies in how we measure traffic.

    At Chartbeat, we measure “concurrents” — the number of users on a page at a given second. That’s fundamentally different from traditional pageviews because the number goes up depending on how long users stay on the page: for instance, if 100 users come to a page each second and they all read for a minute, then after a minute there will be 6000 concurrents on the page. But, if instead they all bounce after 1 second then there will only be 100 concurrents on the page (see http://chart.bt/uWcvYp if that explanation didn’t make total sense). I suspect that differences in time-on-page between desktop and mobile explain the discrepancy between established norms for percentage of mobile traffic and our 4% figure.

    When we state that 4% of our data comes from mobile, what we’re saying is that, at any given second, about 4% of people on a page are on mobile devices. Similarly, when we say that 17% of users have dark social as a referrer, we’re saying that, considering users on at any given second, 17% of referrals came from dark social. That doesn’t mean that 4% of clicks are on mobile or that 17% of clicks are from dark social, because in both cases clicks and concurrents measure different things.

    Going back to the issue of mobile, then, what we’re saying is that even if every one of those 4% of people was referred from an app, it couldn’t possibly have affected the dark social number by more than that same 4%, so Alexis’ observation that “true” dark social is a substantial driver of traffic is correct based on our data.

    Thanks for the critique, though, and for raising the point about mobile analytics — it’s definitely something we should all be thinking about for the future.

  2. avatar Josh Davis says:

    Thanks for the response Josh. I appreciate you providing context.

    Your explanation points both to the issues of bounce rate & engagement time of mobile users as well as the idea that if a share is made in a more personal way, such as email, that the user may be more likely to stay engaged with the content longer. Certainly makes me think.

    Since you say mobile is based on OS, I assume that means that iPads are tracked as mobile traffic?