An Analysis Unlocks Best Practices In Online Content Publishing

While we all might follow many of the best practices in terms of SEO and readability, it’s worth taking a look at what you’re writing to understand what is tracking well, versus not. You might change your mind about some conventions or alter your future publishing habits.

A colleague, Davis Weidman, and I were recently onsite at a client in Minnesota and I was amazed at his analytic prowess. I liked how he unpacked the data and looked for patterns. I was simultaneously working on a new Forbes.com column, and Davis offered to take a look at the performance data of my past work, nearly 100 posts since 2012. Free, expert consulting? You bet.

What does good look like

Quality can be measured in many ways online. Awards won. Retweets. The number of backlinks. But the generally accepted opinion is that more views, unique views, are better. That’s how most of the industry is set up. Weidman first wanted to see which variables, if any, may be driving views for the posts.

He was clear that there were important variables he could not evaluate. Some of these variables likely have an influence on the degree to which a post catches someone’s eye and elicits a click. These might include:

• The time of day that it was posted
• Thumbnails and picture previews displayed for each post

There are also variables that drive share reach (and thus multiply the potential views) he wouldn’t be able to check, such as:

• The specific people that “liked,” shared, and/or retweeted the post.
• How “good” the content was. Quality not measured by popularity.

The variables he could examine: 

• The content — I categorize my posts into four types: Strategies, Careers, Observations or Humor. I also track if the article includes a framework, which can both demonstrate and illustrate one’s thinking. Consultants love frameworks. Maybe people love frameworks.
• The timing — When it was posted: the day of the week, the general season and specific month, and the year.
• The title — How it was worded: did it include a brand name (e.g. Uber, Yahoo); was it a listicle (5 Ways…); was it a question; what were the key words (e.g. “Creative” / “Creativity”)

Weidman ran a series of multiple regression tests across different samples that had varying levels of outlier strictness. To complement this analysis, he also looked at simple averages of views for each post across each of the variables. He mentioned how, despite my best efforts, this is still a small sample size, especially compared to the number of potential variables that we can (and can’t) test, and that as a result the data should be taken with a grain or three of salt. And while some of the variables proved (thus far) to be statistically insignificant, others have raised some interesting points to keep in mind. All that said, here are his findings:

Type of post:

• There was little correlation between types of posts (e.g. humor, observations) and views in the regression. However, careers-related posts had a higher view average than other categories that I’ve been representing more heavily, which may be enough for me to think that I should give more representation to the careers category. People may be more likely to click on articles that may help them navigate their own career; this makes sense in a culture craving for mentorship, shorter employment cycles and also in the Forbes.com context.
• Posts with frameworks do have a high view average and view correlation. This idea may tie back to some of the “untested” variables mentioned above–posts that include frameworks may also inherently be “higher quality” posts while eliciting others with large audiences–e.g. consulting leaders–to share through their own channels.

Timing of post: 

• There appears to be a general lull in the summer and a sharp rise in the fall, which hadn’t occurred to me (unfortunately, I haven’t posted enough in December or January to properly measure the New Year effect!).
• As it turns out, posting on Saturdays had a moderately high positive correlation with views, while Sundays had a moderately high negative correlation. I was surprised at the idea that Saturday is a good day to post, since the conventional wisdom at Forbes.com and many blogs is that Monday at 10 a.m. is magic hour.
• What was I doing right in 2017, I wonder? Posts from this year fared extremely well, with a high positive correlation and view averages that were far and away greater than that of posts from other years. Other years weren’t highly correlated to views.

Wording of the title:

• Titles phrased as questions had a strong positive correlation with views.
• Titles with “digital” (14% of my posts) and “content” (6% of my posts) both had high view averages and very strong. Curious how “Narcissism” will fare.
• Other variables like the inclusion of brand names, listicle formatting, and other key words didn’t prove to be meaningful.

The actionable recommendations:

1. Use titles with “digital” and “content” more often. These key words could be call-outs to audiences who are looking to consume this type of content specifically.
2. Frame titles as questions more often (currently 11% of your posts). Readers may find these intriguing and click hoping to find a specific answer.
3. Take a look at my posts from 2017, and look for any obvious themes or components that you could apply now–you crushed it that year.
4. Try posting more on Saturdays. While it’s different from what I expected, Saturday posts did so well I should try to tease this out more.
5. Careers-themed posts are currently 24% of my total (compared to 38% and 32% for strategies and observations, respectively). Balance this out a bit more to include more Career posts, which have high averages and may be most appealing to people already on Forbes or LinkedIn. This is a need in the market. Humor, sadly, less so.
6. In the summer, post things that are extremely topical or relevant to what’s happening in that moment (versus generic posts that could live in any month), as summer postings appear to dip in views and high quality, season-agnostic posts may not get the reach they deserve. On the contrary, increase the number of posts in the fall , which has a high average view count and is a time when people may be thinking about reaching a goal by end of year. Save the money makers for the end of the year.  

Due to both the small sample size and the likely importance of variables that weren’t tested, Weidman doesn’t believe the data is perfect; however, it’s certainly a good way to highlight and refine some potential hypotheses to explore in future posts and help me continually improve as a writer, an opinion leader, and a contributor.

How well does this post do? Well, for starters, it was posted on Saturday.