Download - The big marketing blog popularity study
The big marketing blog popularity study22 graphs comparing how different kinds of posts lead to more, or fewer, shares/tweets, at Econsultany.com, the Hubspot blog and the SEOMoz blog
Conceived & designed byRyan Skinner (@rskin11)
BACKGROUND• Data was collected from publicly available tweet
counts on-site, in April/May 2013
• Sample sizes range from 7 to 40. Each site’sbaseline was established from a sample size of 40posts. Most common sample size = 10 posts.
• Circle sizes and position on Y-axis correspond to average number of tweets per post in sample.
• Bar = first quartile to third quartile of sampleWhiskers = entire range of sample
• Open data: Link to Google Docs Spreadsheet
“First zegraph, zenze analysis”
FIRST UP, ECONSULTANCY’S EXCELLENT MARKETING BLOG
A note on the Econsultancy data:Econsultancy’s blog post tweet counts
display only tweets made by US readers. Thanks to G. Charlton at Econsultancy, I was able to get the tweet counts from UK visitors,
as well. (Thanks, Graham!)
Question #1:Do Econsultancy’s leading staff
bloggers get significantly different amounts of tweets per post?
Econsultancy.com tweets per post by post author
200
400
600
800
10001188 1702
RyanSommer
DavidMoth
GrahamCharlton
AshleyFriedlein
ChrisLake
There are no losers here. Sommer averages 100+ tweets per post (unattainable for us mere mortals).
Sommer’s and Moth’s averages are brought down due to their contributions to series like Q&A and start-up posts, which garner few tweets.
As CEO, Friedlein’s infrequent postings get a lot of tweets (*but not universally)
Lake’s talent at making research posts on recent topics that are easy to digest get huge tweet counts (watch trend: Accessible, research-heavy content wins!)
Friedlein’s and Lake’s big outlier posts (1000+) tweets lift their average into a different league
Theory of the slugger:All bloggers have their “batting average” but sluggers are able to reach well above this fairly frequently – outlier posts of very
high popularity.In this instance, Lake’s the slugger.
Question #2:Do Econsultancy’s guest
bloggers get more or fewer tweets than the staff bloggers?
Econsultancy.com guests vs. baseline tweets
200
300
400
500
600648
80
Baseline Guest posts
50
Interestingly, guest posts track the Econsultancy baseline (average tweet count over a sample of 40 posts) almost perfectly.
In other words, guest bloggers do no worse at drawing tweets than staff bloggers, on average. However, there are variations from guest blogger to guest blogger. How do they vary, depending on the title of the guest blogger, for example? We’ll see…
Question #3:How do different job roles/titles
among Econsultancy guest bloggers relate to tweets per post?
Econsultancy.com guest post tweets by author title
100
200
300
400
500
CEO/FounderHead of
<Discipline>
Guests
Marketing titles
Consultants
One might expect CEOs to draw more attention, but the data showed them tracking a bit under the average among guest bloggers. Presumably, they find it harder to resist writing sales-heavy, me-centric posts.
“Consultants” is a vague title, but these did noticeably better than other titles (though, it must be noted from a small sample).
Marketers (CMOs, VP Marketing, Marketing Manager) did better than their peers on average. We might infer they know their audience well, and might have more experience with blogging than average (thus getting more tweets).
Question #4:How do different blog topics
(search, social, mobile, etc.) relate to tweets per post?
Econsultancy.com post tweets by topic
100
200
300
400
500
Start-UpPosts
PostBaseline
“Content”Posts “Mobile”
Posts
648 881
Q&APosts
“Social”Posts
“Search”Posts
First of all, the posts profiling tech start-ups and Q&As get far fewer tweets per post than the blog’s average. Clearly, for this blog’s audience, these are of niche interest. Few readers have a motive to share these posts.
At the other end of the scale, search and social were well above the blog’s average. Search beats social on average, but is heavily influenced by one outlier. Posts on social media varied considerably (broad range).
Posts on content marketing and mobile marketing tracked just above the blog’s average tweets-per-post.
Question #5:Do posts about specific tech
brands/vendors get more love in terms of tweets per post?
Econsultancy.com guest post tweets by brand topic
50
100
150
200
250
MicrosoftPosts Apple
Posts
FacebookPosts
PostBaseline
287 366 648
It’s worth noting that sample posts from all three brands generated fewer tweets on average than the Econsultancy blog baseline.
Posts about Facebook, however, generated a lot more buzz than most posts about Apple or Microsoft (each had one outlier, about Apple’s unwillingness to engage in social media and Microsoft’s use of social media, respectively).
Most posts about Facebook, Microsoft, Apple or any one particular tech or brand are reporting by Econsultancy, not best practice or strategy. In general, these reporting posts seem to get fewer tweets.
Theory of the human filter:People prefer to share/tweet posts that have already been analyzed
and pre-digested (compared to first-hand reporting).
Question #6:Do posts that aim for controversy
get more love in terms of tweets per post?
Econsultancy.com controversial posts vs. baseline
100
200
300
400
500691648
PostBaseline
ControversialPosts
ControversialPosts
(sans outlier)
For reference, posts courting controversy included “Is the RFP Broken?” and “Is this social marketing framework accurate (or even necessary)?”
The controversial posts got on average slightly more tweets, but discounting one outlier (“Coca-Cola: The S in social media doesn’t stand for sales”) they actually averaged well below the baseline of tweets for Econsultancy posts.[It’s worth noting that the baseline includes an outlier or two as well, but these are averaged down by the large sample size].
You can’t draw hard and fast conclusions from a small sample, but courting controversy does not seem to automatically equal “more shares”.
Question #7:Do posts with a negative premise
get more love in terms of tweets per post?
Econsultancy.com negative posts vs. baseline
100
200
300
400
5001704648
PostBaseline
NegativePosts
NegativePosts
(sans outlier)
For reference, posts with a negative premise included “Five pointless discussion posts” and “Why a Chief Digital Officer is a bad idea”.
Negative posts got on average far more tweets – almost double, but discounting one outlier (“14 lousy web design trends that are making a comeback”) they averaged only slightly above the baseline of tweets for Econsultancy posts.[It’s worth noting that the baseline includes an outlier or two as well, but these are averaged down by the large sample size].
Conclusion:Despite a clear boost, it is heavily influenced by one big outlier. The greater average of tweets for negative posts sans outlier probably isn’t significant enough to warrant a “how can we give the post a negative spin?” approach.
Question #8:Do posts with an ambiguous title (doesn’t explicitly state content of
post) get more love in terms of tweets per post?
Econsultancy.com tweets: ambiguous title vs. baseline
100
200
300
400
500648
BaselineAmbiguous
Titles
A personal theory: Many people like to write clever, slightly poetic titles in the belief that some wit will win more shares. This was meant to test that.
Conclusion:When a post’s title doesn’t very specifically state what the post will be about (ex. “Why you need real options” and “Are you getting personal with your customers”), it seems to get fewer shares. In other words, if people are writing poetic titles to get more shares, they would more often than not be…wrong.
(for clarity’s sake AND for twitter’s sake, give your post a clear and descriptive title.)
Question #9:Do posts with lists get more love in
terms of tweets per post?
Econsultancy.com tweets: list posts vs. baseline
100
200
300
400
500648
Baseline
List posts
535
The data’s pretty clear on this one. Posts like “Seven tips for charities using Twitter” and “11 great ways to use social proof in ecommerce” are shared more often on average.
Econsultancy’s own data also shows they’re viewed more often. Same for guest posts with lists. People love their damn lists.
Generally, you need to do some amount of original research to create a list of examples or approaches. And people just love to share original research.
Conclusion:If you don’t like lists, get over it. They’re useful, people like them and people share them.
Question #10:Do posts with a title as question get
more love in terms of tweets per post?
Econsultancy.com tweets: question title posts vs. baseline
100
200
300
400
500648
BaselineQuestion title
posts
Do posts with titles as questions get more shares, on average? The answer is no (at least not in a significant-enough way to move the dials).
In terms of popularity, posts posing a question don’t get shared more often than those that don’t.
Question #11:Do posts with lots of images (8 or more per post) get more love in
terms of tweets per post?
Econsultancy.com tweets: image heavy posts vs. baseline
100
200
300
400
500648
BaselineImage heavy
posts
We’re often admonished to “show, don’t tell”, and told of the power of the image. This may be true, but – based on this sample anyway – there’s no correlation between producing an image-rich post and getting more shares on twitter.
This doesn’t mean to say putting images in a post is a bad thing (on the contrary) but they do not automatically make your post more shareable.
NEXT UP, HUBSPOT’S GREAT IN-HOUSE MARKETING BLOG
A note on the Hubspot data:Hubspot’s blog seems to have seen a marked
increase in traffic and tweets recently, which could skew results. For that reason, the baseline (taken
from more recent posts) could be artificially inflated. Otherwise, comparisons are like for like (old to old,
new to new).
Question #1:Do posts for the introductory,
intermediate or advanced digital marketer get more tweets than
Hubspot’s baseline?
Hubspot blog tweets by post difficulty level
500
1000
1500
2000
2500
IntermediatePosts
BaselineAdvanced
Posts
IntroductoryPosts
2976 4182
Hubspot classifies their own posts according to difficulty level: Introductory, intermediate or advanced. Which tend to get more tweets?
The only significant uplift seems to come from the posts they label introductory. You can draw any conclusion from that you like. Mine: Introductory posts are more readily available to more marketers, and thus get more tweets. (A smaller proportion of their readers will be interested in the most advanced topics).
Theory of the mass mass:Posts that target the broadest swathe possible of a given blog’s typical readership will, unsurprisingly,
often garner more tweets than narrow posts. However, earning tweets beyond “typical readers” is
another story (and may be more important to readership growth).
Question #2:Do posts with trigger words like
“awesome”, “love” or “simple” get more tweets than average?
Hubspot blog tweets by trigger words
300
600
900
1200
1500
“Fantastic”Posts
“Love”Posts “Simple”
Posts
4183
“Awesome”Posts
Baseline “You”Posts
4182
Drawing a connection between one trigger word in a post title and the number of tweets that post gets, is tenuous at best, admittedly. So the results should be taken with a grain of salt.
Interestingly, however, self-described “awesome”, “fantastic” and “love” posts got significantly fewer tweets on average (overcompensation, anyone?).
Posts that address the reader specifically (“When you should and shouldn’t outsource your marketing”, for example) seem to do better at winning tweets from readers. Note, however, that this result is strongly influenced by one outlier. Sans outlier, it would probably lie even with the baseline.
Question #3:Do certain “types” of posts (lists,
negative sentiment or how-to’s) get more tweets than average?
Hubspot blog tweets by post type
300
600
900
1200
1500
NegativePosts
(sans outlier)
How-ToPosts List
PostsBaseline Negative
Posts
4182 4193
Some similarities and differences from the Econsultancy data here. Negative posts seem to do better, but – sans one outlier – they do significantly worse.
It’s not clear why list posts would do, on average, worse than the blog’s baseline for tweets per post, unless it’s because the baseline is inflated towards the recent upwards bump in traffic and shares.
Discounting the baseline and the single negative post outlier, we see that list posts get marginally more tweets than how-to’s, and these get marginally more tweets than negative posts. This probably corresponds to their relative utility seen in the eyes of Hubspot’s blog visitors.
Question #4:Which get more tweets: Marketing
posts about how to optimize LinkedIn, Twitter or Facebook?
Hubspot blog tweets by social network as topic
500
1000
1500
2000
2500
“Twitter”Posts
“Facebook”Posts
“LinkedIn”Posts
4252
Hubspot loves to do posts helping marketers optimize their activities on the major social networks, and particularly “the three biggies” – LinkedIn, Twitter and Facebook.
Which garner the most tweets? Surprisingly, not twitter. The differences aren’t huge, but they are significant. Posts about Facebook get shared the most, and those about LinkedIn get shared least.
(It would be interesting to study the same sample on LinkedIn shares and Facebook likes or shares, but that’s another study).
Question #5:Which Hubspot blogger gets the
most tweets on average?
Hubspot blog tweets by Hubspot blogger
300
600
900
1200
1500
BrittanyLeaning
PamelaVaughan
MikeVolpe
CoreyEridon
DanZarrella
1654 2216
Like with Econsultancy’s Ashley Friedlein, one might expect Hubspot CMO Mike Volpe to get the biggest tweet cull. However, he’s edged out in the sample by Corey Eridon, and blown away by Dan Zarella.
It’s particularly worthwhile ignoring averages on this one and looking at the bars (first to third quartiles), as Volpe’s average is tweaked upwards by a hefty outlier.
For Zarella, even though every post doesn’t generate huge numbers of tweets (one got only 193 – gasp!), he makes very shareable posts, very consistently.
Theory of the slugger reconfirmed:Like Econsultancy’s Lake, Zarrella’s got a research and data focus, combined with a
knack for storytelling.As a result, he consistently nails highly
shareable posts.
FINALLY, SEOMOZ’S HUGELY POPULAR DIGITAL MARKETING BLOG
A note on the SEOMoz data:Nothing particular to mention here. Tweets
were culled from SEOMoz’s own tweet counts, and post samples were drawn as randomly as possible over the past six
months, approximately.
Question #1:Which gets more tweets on the SEOMoz blog: Technical SEO posts or social media posts?
SEOMoz blog tweets: Technical SEO vs. Social Media
300
600
900
1200
1500
Baseline
Social MediaPostsTechnical SEO
Posts
2577 1664
At first pass, this came as a surprise. A community of SEO’s more eager to share posts about social media than technical SEO – what?!?!
However, it’s worth noting that a social media post on SEOMoz is often about how social media works in an SEO context. Almost any SEO will be interested in that. And a technical SEO post on SEOMoz is for the SEO with a deeeeep interest in technical SEO – a smaller group, even at SEOMoz.
Theory of the mass mass reconfirmed:Again, posts that appeal to the broadest swathe of your typical audience will, because a (almost) fixed proportion of interested readers tweet any given
post, get the most tweets. (These posts may not be those that win you the new readers, though. In fact,
they often don’t.)
Question #2:Which gets more tweets on the
SEOMoz blog: Content & blogging posts or Vertical SEO posts?
SEOMoz blog tweets: Vertical SEO vs. Content/Blogging
300
600
900
1200
1500
Baseline
Content & BloggingPosts
Vertical SEOPosts
2577 20801635
Similar result to the first head-to-head. Again, the “non-SEO topic” walked off with far more tweets than the SEO topic.
Like the first study, however, I think vertical SEO interests a narrower sub-set of SEOs than the broader content topic, giving the latter more tweets. (I would also expect that here, as in the previous study, the non-SEO topics drew in more tweets from people who don’t typically read SEOMoz’s blog).
Question #3:Which gets more tweets on the
SEOMoz blog: Analytics posts or Link Building posts?
SEOMoz blog tweets: Analytics vs. Link Building
500
1000
1500
2000
2500
Baseline
Link BuildingPosts
AnalyticsPosts
2577 4022
Link Building posts pulled in, on average, far more tweets than analytics posts. I’d love for any SEOs to weigh in here, but I suspect that link building’s a bit like fishing – they’re always thinking about how to do it better, trying to get the big one and feeling a bit frustrated by it.
Analytics posts, like vertical SEO, appeal to a narrow subset of the blog’s readers, leading to fewer tweets.
Question #4:Which get more tweets – all topics?
SEOMoz blog tweets by topic
500
1000
1500
2000
2500
TechnicalSEOPosts
AnalyticsPosts
SocialMediaPosts
4022
VerticalSEOPosts
ContentPosts
Link BuildingPosts
Seen on a line, we see the division between post types for a niche audience (vertical SEO, technical SEO and analytics) and those for SEOMoz’s broad audience (social media, content and link building). This is particularly revealing when you look only at the bars – first to third quartile – and ignore the averages and outliers (whiskers).
Question #5:Which get more tweets – posts with few comments (< 50) or posts with
more comments (> 50)?
SEOMoz blog tweets: Less comments vs. more
300
600
900
1200
1500
Posts with> 50 comments
Posts with< 50 comments
Comments are often a proxy for a post’s credibility or simple “interesting-ness”, and the assumption was that this would correlate to average tweet counts.
The results loosely corroborate this finding. All in all, not a surprising conclusion.
Question #6:Who gets more tweets – Rand Fishkin, Dr. Pete or YouMoz-
promoted posts?
SEOMoz blog tweets: By writer
500
1000
1500
2000
2500
Dr. Pete’sPosts YouMoz-
PromotedPosts
Rand’sPosts
Again, the head honcho is edged out by a data/research-driven lieutenant – in this case, Dr. Pete. Like Lake and Zarrella, he’s focused on original research, and communicating that clearly to the community.
YouMoz-promoted posts beat both, however, in terms of average tweets per post. This is no surprise, as the SEOMoz team is curating those posts that get the most engagement from the YouMoz blog. While tweets are not their core metric for promotion, posts that get traction are likely to get traction on twitter too.
Theory of the slugger reconfirmed, again:Clearly there’s a genus of blogger represented by
Lake, Zarrella and Dr. Pete, who bring new knowledge to their communities in easy-to-digest ways. This means they get high engagement on
every post (as an average), and tremendous engagement on some posts (grand slams).
Some final observations on the outliers - posts that are kinda viral
(> 500 tweets on Econsultancy or > 1500 tweets on Hubspot or
SEOMoz)
Econsultancy outliers =
Posts that specifically aim to help marketers navigate the context of something they’re trying to
figure out right now, with original research or highly relevant first-hand experience.
Hubspot outliers =
Posts that consist of short-cuts or rich and clear instructions (that aren’t explicitly
“simple”). Trigger words: “Cheat sheet” and “Disaster” or “Catastrophe”
SEOMoz outliers =
Posts that contain an authoritative take for SEOs on a niche topic, such as content
marketing strategy, social sharing, responsive design, facebook or video.
Conclusions
There are clearly patterns worth finding and acting on, but the insights are not obvious or immediate, nor the actions worth taking.
This methodology can be used to discover what works on your own site, or competitors.
Act on hunches. Then measure. And adjust.
Making-of this study:
~ 30 hours (2 hours concept, 10 hours data collection, 8 hours graphs, 8 hours analysis, 2 hours editing)
Built on methodology first proposed here: http://www.velocitypartners.co.uk/our-blog/content-marketing-measurement-one-metric/
(Please share if you enjoyed it :)