You asked us to extend the analysis we did for OkDork posts from 2014 and here it is! Neville Medhora from KopywritingKourse gave us the data from six popular marketing blogs with more than 100,000 followers.
- OkDork blog by Noah Kagan,
- Backlinko blog by Brian Dean,
- ConversionXL blog by Peep Laja (no post date and no number of comments),
- IWillTeachYouToBeRich blog by Ramit Sethi (no number of comments),
- Quicksprout blog by Neil Patel,
- SocialTriggers blog by Derek Halpern.
The data contained basic information about post url, date posted, social share counts, headline, length of headline. We ran textual analytics on the post content and the headlines and added many additional metrics like post neutrality, headline neutrality, frequencies of most common words, weekday when posted, number of external links per post, etc. - 52 metrics in total.
The data here be found here: Blog Data
The question is: "Can we find out what's the perfect post recipe based on the data from marketing & blogging giants?"
The answer is as always right in the data! We did quite a bit of analysis with this data and created something cool for you to play with. Here is a structure of this fat and juicy post:
- Basic statistics. Comparing blogs with each other using main metrics.
- Trends for individual giants over time.
- Headline analysis. Explore best headlines by words used.
- Content analytics. Five post segments emerged from the textual analytics.
- Sentiment analytics. We analyzed sentiments of posts themselves and post headlines and made interesting discoveries.
Here is a summary of our findings:
- It's good to be consistent. All blogging giants with a large following consistently make between 10 and 40 posts per calendar quarter, i.e. between one and 3 posts per week. (Per blog the variance is very small).
- It's not true that there is a magic day to publish posts. For different blogs the best week day changes. The only certainty is that weekends work worse for social shares than workdays.
- It is not good to be neutral. Sentiment analysis of headlines and post content shows that neutrality is a negative driver for social shares and comments. Positive sentiment of the post content and headline drives most social shares. [Please, use with caution. Out data contained highly curated content from well respected authors.]
- If all posts are analyzed together the optimal post length to maximize tweets goes as far as 4000 words.
- Six distinct post groups emerge from all the posts we analyzed, and two of them have significantly higher number of shares per post. These groups have the following combination of core words: (1) test, page, conversion, customer, use, people, product, make, site, user, good, one, design, like, email, image, website, visitor, rate, need, and (2) content, link, site, traffic, social, search, use, blog, post, google, page, rank, image, people, infographic, create, like, make, write, share.
- Links to external websites are an important driver of popularity. For small posts of under 1000 words 8 to 10 external links are the optimum. Large posts require at least 50 links to hit the maximum sharing.
Blogs by the Numbers
We wanted to call this post: "Stand on the shoulders of marketing giants" for a reason. Six blogs have given us a list of 2,381 posts with whopping stats:
- 2,577,139 words
- 372,626 twitter shares
- 119,281 facebook shares and 135,095 facebook likes
- posted over 1795 days between 2004 and 2015
- 114,609 post comments.
People who write this goodness are machines! Many of us wonder how to get large following - we're afraid there are few shortcuts.
Look at the TWEETS of individual blog posts over time.
Here is a zoomed version for all posts with up to 500 TWEETS. Each dot is a carefully crafted story!
All possible headline lengths have been tried. Headlines with 40-65 words have high fraction of super-tweets:
In a zoomed version we see that different blogs have different preferences and the high-popularity range of headline lengths is quite wide.
Word count varies from blog to blog. Backlinko, OkDork, IWT like long and super-long posts of more than 2000 words:
Socialtriggers prefers posts with less than 1000 words, while Quicksprout has specific preferences for either short 250-300 word posts or 1000-2000 words:
Let's look at posting evolution for all giants over time
For up to date information we only selected the posts published since January 2014 and did the analysis of main blog metrics on quarterly basis. Normally calculating average values per quarter would depend on the "outliers", e.g. posts which went viral. That's why we also did the robust stability analysis which led to same conclusions. If you want to see the results - contact us.
Look at the differences in posting frequency per quarter since 2014. OkDork has been getting quieter since Q3 2014. Quicksprout outputs 35-40 posts per quarter. Ramit Sethi has been ramping up his output significantly since Jan 2015.
Average FACEBOOK SHARES per POST are quite different among the blogs. We clearly see that social sharing is not everyone's focus and there are good reasons for that. Note the backlinko, for whom we've got data since Q2 2015 has a quickly growing trend in avg shares per post:
Note, that we did not have a complete information for Q3 2015 which is just over, and that's why we don't plot the metrics for other blogs except backlinko. The amazing thing about backlinko is that we only 29 posts in our dataset the blog has very impressive social sharing metrics. If shares are the goal, and time is tight Brian Dean from backlinko might have a process which falls into the growth hacking category.
Average TWITTER SHARES per POST show similar trends. Note, that Q3 2015 might have incomplete information. Can you guess which blogs have social shares as their target metrics?
Average HEADLINE LENGTH per POST quite similar for all blogs - something between 40 and 60 characters is a golden range:
Average WORD COUNT per POST differs a lot. Socialtriggers have been ramping it up. Backlinko started writing super-long posts.
Weekday to post to get shares ranges significantly among blogs. The only general conclusion is to not post on weekend.
We analysed the data with respect to specific patterns in the weekday of posting and social shares. If all 2,831 posts are considered as one pile - the results are inconclusive.