If you want to grow your audience, you need to understand how people share your content.

At Percolate, our readership comes primarily from direct traffic and Twitter referrals. Direct traffic is also known as “dark social” because it’s likely that most of these readers are not actually typing in our URL, but clicking to it from links in emails, as well as chat and mobile applications which don’t result in a referral source. [1] Traffic from Twitter is more straightforward: links to our blog posts get shared by @Percolate and lots of other folks on Twitter, and people click on them.

It follows that if we want to grow our traffic, we should try to understand why our audience chooses to share things over these channels. Which brings us to a question: is there a difference between content shared on Twitter vs Email?

A quick look into our own Google analytics wasn’t particularly insightful: the top articles for both traffic sources were similar. Perhaps this was a dead end. But then, something interesting happened.

I was reading the New York Times and browsing their “Most Emailed” and “Most Tweeted” lists and noticed that there were significant differences in what was shared via each medium. There were overlaps, sure, but the orders usually varied and a number of articles showed up on one list but were completely absent on the other.

This called for further investigation

Working together with Monik Sheth, a member of our business insights team, we spent an afternoon looking at some data. If you want to skip straight to the findings, click here. If you want to see our somewhat ad-hoc analysis, read on.

How We Looked at the Data

We put together a spreadsheet with 4 tabs on March 19th 2014 with the following data:

  • Top 20 Most Shared via Email (Last 24 hrs)
  • Top 20 Most Shared via Email (Last 30 days)
  • Top 20 Most Shared via Twitter (Last 24 hrs)
  • Top 20 Most Shared via Twitter (Last 30 days)

Check out the spreadsheets on Google Docs

The goal here was specifically to look the differences between the lists. We compared the two article lists for overlap using a function called VLOOKUP to show what rank a specific article had in the other list, if it appeared in the list at all. [2] We compared the ranks of top emailed and tweeted articles in the last 24 hours and the last 30 days against each other. We then subtracted the rank of the article in one list from the rank in the other list and get a distance or “gap factor.” (Ranked 2nd in one list but 18th in another list produced a “gap” of 16.)

We were originally going to base this post on articles with a big gap factor, but then we realized that if the title didn’t show up at all in the other list, it was an even more significant indication. In other words, we believe that top articles only shared significantly via one channel revealed the most about people’s sharing preferences.[3]

The Article List





Apple’s Serious Security Issue: Update Your iPhone or iPad Immediately How to Get a Job at Google Silicon Valley’s Youth Problem
The Science of Older and Wiser Reaching My Autistic Son Through Disney Obesity Rate for Young Children Plummets 43% in a Decade
Learning to Cut the Sugar What You Learn in Your 40s Ukraine Mobilizes Reserve Troops, Threatening War
Paul Ryan’s Irish Amnesia When May I Shoot a Student? How Sotnikova Beat Kim, Move by Move
The Story Behind the SAT Overhaul The Fat Drug Billionaires With Big Ideas Are Privatizing American Science
Your Ancestors, Your Fate Space Ripples Reveal Big Bang’s Smoking Gun Piers Morgan and CNN Plan End to His Prime-Time Show
Study Questions Fat and Heart Disease Link A New SAT Aims to Realign With Schoolwork Credit Unions Offer Creative Home Loans
Get Out Your Corkscrew Spain, Land of 10 P.M. Dinners, Asks if It’s Time to Reset Clock London’s Laundry Business
How to Get Fit in a Few Minutes a Week Study Questions Fat and Heart Disease Link A Story of Perseverance
Gadgets to Boost Bike Safety Scientists Sound Alarm on Climate Ukraine Crisis in Maps
To Keep Teenagers Alert, Schools Let Them Sleep In Helping Women Get Back in the Game El Chapo, Most-Wanted Drug Lord, Is Captured in Mexico
The Sound of Philadelphia Fades Out Out of Control Where Are the People of Color in Children’s Books?
Sorting Out the Risks of Fish From Putin, a Blessing in Disguise Lost Jet’s Path Seen as Altered via Computer
The Unlikely Road to War Reading, Writing and Renewal (the Urban Kind) What Pakistan Knew About Bin Laden
How a Warm-Up Routine Can Save Your Knees The Rise of Anti-Capitalism If Not a Cold War, a Return to a Chilly Rivalry
Clarissa Dickson Wright, Rebel TV Chef in ‘Two Fat Ladies,’ Dies at 66 Putin Reclaims Crimea for Russia and Bitterly Denounces the West Three Years of Strife and Cruelty Puts Syria in Free Fall
A Tumor, the Embryo’s Evil Twin If History Is a Guide, Crimeans’ Celebration May Be Short-Lived Budding Liberal Protest Movements Begin to Take Root in South
Is Crimea the Next Yugoslavia? Google Unveils Software for Smartwatches-to-Be
Tesla Fights for a Place to Park Pastry Chef to Obamas Hanging Up His Whisk
A Campaign Inquiry in Utah Is the Watchdogs’ Worst Case Airlines Use Digital Technology to Get Even More Personal
White House to Introduce Climate Data Website



In total we collected 20 items off of four lists for a total of 80 “article slots”. A few of the articles that were most shared on the Last 24 Hours list really moved the needle; they also appeared in the Last 30 Days list. We were left with 77 articles. About half of the articles on each channel (Email or Twitter) Top 20 List did NOT appear in the other channel’s Top 20 List , meaning we had 58 unique articles out of 80 possible slots.

Essentially there was a 40%-60% overlap between the articles most shared on Twitter vs Email.


Articles shared over email often relate more to people’s personal lives and interests. You can easily imagine people sharing articles with their Apple-wielding friends around the iOS security issue. Another example might be an article about how cancer tumors mirror the activity of a growing embryo, sparking a heart-felt discussion with friends about the disease. Finally, you can imagine lots of young adults forwarding their parents vindictive proof that, no really, you should have let them sleep in.

On the other side, links shared on Twitter seem to focus on the broader world, particularly international affairs, which the most common topic in the Most Shared on Twitter articles, taking 7 of the 20. Everything from Syria to Ukraine to Pakistan to Mexico was covered in the list.

As for articles that overlapped, you could see how the article might apply personally and yet tie into a bigger theme. Many people may have emailed How to Get a Job at Google as advice to a friend, ex-coworker or child. Meanwhile, others may have tweeted it as so many people are looking for work and it touches upon the declining relevance of traditional education. Additionally, the aggressive moves of President Vladimir Putin around Crimea broke through people’s personal radars. It was one of the international affairs articles that became a hot discussion topic over email as well as on twitter.

We created a diagram for these overlapping spheres:



When it came to email, health-related articles were most common. 8 of the 20 articles had a health bent,  including: an article about cutting out sugar; another about how certain exercises could protect ones knees; and another dealing with the benefits and drawbacks of fish consumption. Other posts were related to aging or just the aspects of life for older people: a reflection on getting older and wiser; a story about how a certain American regional accent is going away; and the obituary of a popular television chef in the UK.


Meanwhile, beyond international affairs, Twitter articles had more of business and technology focus. One article touched on Google’s plans to offer operating systems for smartphones; another discussed airlines growing usage of more high-tech gadgetry in the skies; and yet another covered the trend of credit unions moving into the home loan market.

Final Takeaways

Ultimately, we’re working with a really small dataset here and this comes out of just a few hours of thinking and spreadsheet jockeying. Stepping away from the data for a bit, I’d like to share some general impressions and ideas I had after looking at the articles:

Things that are shared over email tend to be thoughtful and have a more personal relevance. People are usually sharing with a smaller group of trusted individuals. Perhaps they are attaching commentary or expect to go into a longer discussion about the article. These folks might be a little older, with children. We might deem this category of content sharing “My World”.

Meanwhile, on Twitter, people are aware of the fact that their broadcasts are public and can be seen by anyone. They probably want to make sure their content is interesting to the diverse followers they have. They may be more interested in what’s happening RIGHT NOW, which is why they love the real-time social network. They probably care about things far outside of themselves and are interested in technology. We might call this “The World”.

You’re welcome to draw your own conclusions from the data. But they still make for interesting discussion. If we were to restate Marshall McLuhan a bit: the medium shapes what messages get shared.


[1] Dark social was reported on in late 2012 by Atlantic Magazine. Other analysis by Vikrum Nijjar indicate that it’s not as clear as anyone may think. For now, I’m using email as a proxy for dark social.

[2] VLOOKUP stands for Vertical lookup – it’s a function in Excel and Google Spreadsheets that Monik really helped me figure out. The general syntax is : VLOOKUP(search_key, range, index, [is_sorted]). It searches down the first column of a range for a key and returns the value of a specified cell in the row found. You can learn more here.

[3] Yes, it is possible that the #18 article in “Most Tweeted” might actually be #25 in “Most Emailed” and since we only have access to the Top 20 in each category, we would never know. But we feel confident that with so many articles on NYTimes.com, the chances of this actually being the case is low. Our assumption is that if an article didn’t show up in the top 20 on one list, they were likely way farther down.