The Like Log Study is released under Creative Commons Attribution 3.0 license by Yahoo! Labs.
This is an independent copy of the study orginally published at http://ediscope.labs.yahoo.net.


[Single Infographics]


Reference: Yury Lifshits. Ediscope: Social Analytics for Online News. Yahoo! Labs, Tech. Report No. YL-2010-008.

Short URL: http://likelog.yury.name
Contact: yury@yury.name
Released under CC BY 3.0 license

The Like Log


45 sites, 100K+ articles, 40M+ reactions
Three months from 10/2010 to 01/2011
"Total" facebook counts:
likes + shares + facebook comments

The Rankings

New York Times is the leader of social engagement with 2.3M likes/month, 400 likes for a median story and 13 articles in top 40. Wall Street Journal story "Why Chinese Moms Are Superior" is the greatest hit with 340,000 likes.

The Numbers

There are around 10 likes per 1000 pageviews (across several websites with public PV numbers). Decay of engagement is extremely sharp, with less than 20% likes happening after the first 24 hours.

The Trends

Stories about Facebook, Apple, Verizon, Groupon, future and infographics are universally popular across technology blogs. Articles about Microsoft, Amazon, Samsung, cloud computing, TV and search see much less engagement.

Recommendations

  1. Big effort for big stories
  2. Improve promotion of your best content
  3. Improve your median story
  4. Invest in demand analytics
  5. Invest in social media optimization

The Rankings

Facebook allows anyone to get the total number of likes for any URL. For this study we collected like counts for 45 popular news sites over the period of three months. You can sort our collection by any parameter.
Site Total likes Top story Top 13 stories Top 90 stories Median story # of 3+ liked stories
New York Times 6815796 105289 12% 36% 398 3955
BBC 4331367 140012 13% 38% 225 3867
NPR 2549613 119351 18% 48% 98 2576
The Guardian 1883161 78066 21% 47% 122 2408
Yahoo! Sports 1774183 40490 16% 47% 32 3325
Fox News 1730704 33281 13% 38% 71 3644
Yahoo! News 1693928 42995 26% 60% 95 2476
BuzzFeed 1479737 112289 28% 57% 60 3251
TechCrunch 1441201 7475 3% 12% 484 2595
Yahoo! Shine 1385644 58843 29% 78% 0 751
CNN 1263924 238596 42% 78% 0 690
Wall Street Journal 1205701 342294 58% 81% 1 1353
Mashable! 1041076 69290 23% 44% 190 2035
AOL News 1010507 48115 25% 54% 43 1940
Yahoo! Music 1002684 55338 24% 62% 28 1371
TIME Magazine 890017 328192 51% 72% 28 2709
Yahoo! Movies 758456 42002 34% 80% 45 667
Huffington Post 710561 42005 35% 71% 25 1281
Reuters 655062 152780 42% 67% 15 3281
Engadget 600962 26858 16% 40% 60 2961
The Economist 593271 63317 25% 49% 97 1679
Vanity Fair 592001 44998 46% 80% 35 1027
Forbes 538588 56695 47% 70% 19 2305
Wired 469834 29806 25% 57% 107 1220
Change.org 400062 42310 42% 64% 42 1601
The Daily Beast 370000 14752 22% 49% 17 2140
Bloomberg 352700 15092 29% 57% 11 2906
Yahoo! OMG 317684 23268 38% 73% 6 1929
CNET 305803 11273 23% 54% 31 2028
Gawker 251199 48976 53% 85% 0 563
Yahoo! Finance 220971 30228 52% 97% 0 317
Fast Company 199552 19395 29% 59% 30 1470
Business Insider 199545 13630 46% 76% 3 1239
Good Magazine 197445 7301 22% 51% 25 1706
ReadWriteWeb 144279 13763 39% 60% 15 1676
BNET 136556 13481 44% 70% 5 1540
Business Week 89036 10515 33% 58% 1 1418
eHow 87843 9665 44% 72% 0 945
Inc Magazine 87540 5553 29% 58% 17 953
GigaOm 64357 4284 32% 54% 6 1686
VentureBeat 51021 1804 18% 42% 10 1722
Associated Content 49383 12398 77% 93% 0 405
All Things D 21502 1872 43% 80% 0 401
paidContent.org 9955 942 36% 68% 1 441
Poynter 8660 1207 42% 79% 0 294
We use 13 top stories as a model for "one story per week" strategy and 90 stories as a model for "one story per day" strategy.

Top 40 Articles

Among top stories there are only four articles about factual political news and three about celebrities. The most common type of hit stories is opinion/analysis. Other common themes include: lifestyle, photo galleries, interactives, humor and odd news.
  1. 342294 Why Chinese Mothers Are SuperiorWall Street Journal
  2. 328192 New Zodiac Sign Dates: Earth Rotation Changes Horoscope SignsTIME Magazine
  3. 238596 No, your zodiac sign hasn't changed CNN
  4. 164166 A Holiday Message from Ricky Gervais: Why I'm An AtheistWall Street Journal
  5. 152780 Tired Gay succumbs to Dix in 200 metersReuters
  6. 140012 Woman in 999 snowman theft callBBC
  7. 119351 Arizona Rep. Giffords Shot At Public Event In TucsonNPR
  8. 112289 The 100 Best Signs At The Rally To Restore Sanity And/Or FearBuzzFeed
  9. 105289 Repeal of ‘Don’t Ask, Don’t Tell’ AdvancesNew York Times
  10. 85452 Sisters and Happiness - Understanding the ConnectionNew York Times
  11. 80420 How Obama Saved Capitalism and Lost the MidtermsNew York Times
  12. 78066 Porpoises rescue Dick Van DykeThe Guardian
  13. 77965 Detroit in ruinsThe Guardian
  14. 76432 The Hollywood Issue - 14 Actors Acting - James Franco, Natalie Portman, Matt Damon and MoreNew York Times
  15. 72421 Video Simon's Cat: Santa ClawsThe Guardian
  16. 69290 "FarmVille" Interruption Cited in Baby's MurderMashable!
  17. 68024 Climate of HateNew York Times
  18. 66205 Budget Puzzle: You Fix the BudgetNew York Times
  19. 64045 4-Year-Old Can Be Sued, Judge Rules in Bike CaseNew York Times
  20. 63317 Doctoral degrees: The disposable academicThe Economist
  21. 62806 Hipster Dinosaurs BuzzFeed
  22. 59178 Verizon Finally Lands the iPhoneWall Street Journal
  23. 58923 Our Banana RepublicNew York Times
  24. 58843 Baring It All: The Barefoot Running TrendYahoo! Shine
  25. 57386 Congresswoman Shot at Event in TucsonNew York Times
  26. 56695 Congress Passes Socialized Medicine and Mandates Health Insurance -In 1798Forbes
  27. 55455 Burma releases Aung San Suu KyiBBC
  28. 55338 Ricky Martin: 'I Am a Fortunate Homosexual Man'Yahoo! Music
  29. 54676 During Tonight's Lunar Eclipse, Moon Will Turn Into A 'Reddish Ball'NPR
  30. 54315 Let It Dough!New York Times
  31. 54168 Cartoons Invade Facebook to End Violence Against ChildrenMashable!
  32. 51409 Has Your Horoscope Changed?Yahoo! Shine
  33. 49157 Snow present in 49 of the 50 U.S. statesCNN
  34. 48976 4chan on the Hunt for Puppy-Throwing Girl Gawker
  35. 48115 Lunar Eclipse December 2010 Falls on Winter SolsticeAOL News
  36. 47970 For Law School Graduates, Debts if Not Job OffersNew York Times
  37. 47739 Alcohol 'harms more than heroin'BBC
  38. 45531 From WikiChinaNew York Times
  39. 45284 Mapping America — Census Bureau 2005-9 American Community Survey - New York Times
  40. 44998 Justin Bieber on His Musical Inspirations, His Fans, and Trying to Be a Regular KidVanity Fair

Numbers to Know

Social Engagement Per 1000 Pageviews

Some websites display actual number of pageviews for every article. In this case we can calculate what fraction of visitors is tweeting or sharing a particular story. Our measurements show that for every 1000 visits there are from 5 to 20 likes and tweets. Next time you want to guess pageviews of your competitors, multiply their like counts by 100.

comparison of liking and sharing of stories on Gawker, Business Insider, Forbes blogs, and change.org
Non-top = all articles except top 10.

Lifespan of a Story

Sadly, web articles are basically lost 24 hours after publication. Our measurements show that fewer than 20% of likes come after the first day. Yahoo! News has the sharpest decay of user interest. Engadget articles have the longest lifespan. But even in this case, new content is only visible for 3 or 4 days.
comparison of social activity for average web sites, Engadget, and Yahoo! News
For this experiment we track articles from each of the following sources: Washington Post, Gizmodo, CNN, MSNBC, HuffingtonPost, Yahoo News, New York Times, Engadget, Mashable, and TechCrunch. Note that the total share of activity is significantly less than 100%. This is due to activity in the interval between the time a story was published and the time we have discovered it.

Five Recommendations

  • Put significant effort in your top stories. According to our study, most websites can capture 30% of their total enagement by publishing only ONE story per week! One story per day can capture 70-80% of your audience reactions. But we rarely see web journalists working weeks on a single story. Here are some examples of "big effort for big story": events coverage (Oscars, Superbowl), nominations, awards and voting (Fast Company Influence Project, TIME Magazine Person of the Year readers choice), exclusives (Gizmodo iPhone 4 leak), celebrity Q&A (Obama at Youtube, Beckham at Yahoo! Sports), lists (Forbes Billionaires, Vanity Fair 'Best Dressed'), high-quality infographic & interactives (Good Magazine Transparency blog, NYT 'You Fix the Budget'), trends and predictions (Read Write Web), liveblogging from top events (Apple announcements), op-eds on most controversial issues. Yahoo! Labs have recently published a paper with detailed analysis of hit/niche balance of interests.
  • Improve promotion of your best content. According to our measurements, web stories are practically lost 24 hours after publications. Only 20% likes are coming after the first day. This engagement pattern discourages production of "big stories". To get maximum return on your hits, change your frontpage policy. Best stories should be highly visible. Consider hits-only RSS and twitter feeds, month-in-review / year-in-review programs. TechCrunch Classics is another example of hit promotion. And internal efforts are not enough. Breakout success comes when other media (top TV networks, newspapers and magazines) are picking your story and link back to it.
  • Improve your median story. Sort all your stories by engagement and pick a story right in the middle of the list. This is called a "median". A median story has less than 50 likes for majority of websites in our study. In other words, every second story takes more effort from a writer than it brings value to the readers. Recently leaked "The AOL Way" reports their median story to have only 1500 pageviews, and they aim to grow it four times. Publishers should ask themselves: Why do we write so many weak stories?
  • Use both intuition and algorithms for demand analysis. Some say that best writing can only come from internal compass. Others are writing new articles in direct response to search queries. The key to success is likely somewhere in between. In ideal settings the writers have full control but they are equiped with great tools to navigate content demand. Up to this moment most demand tools were based on search data and internal metrics. The Like Log presented here is a new approach to demand analysis. Likes represent so called "interest graph", the map of world interests. You can understand your readers better by using our engagement trends technology. And you can get insights into audience of your competitors, too.
  • Invest in social media optimization. According to our numbers, you should get around 5-20 tweets and likes per 1000 pageviews. If you have less than that, your social distribution is suboptimal. Create different designs for sharing and subscription functionality and split-test them to find the optimal one. Also, make sure to have just one URL for every story and slideshow. Otherwise, your enagement numbers get scattered among article duplicates. Social networks are likely to be a bigger traffic driver in the future. Combining our data with Comscore, we see that average New York Times reader makes only ONE like per year. We expect this number to go up significantly.

Pointers

Check out the follow ups for this project: How to do a social media check up for your site? and Ideas for like-log-based research projects and web services. Any feedback is highly welcome. Further updates will be announced at @yurylifshits.
Contact Yury at yury@yury.name or @yurylifshits