Over the last few months, I’ve talked to hundreds of writers.
I often asked them:
Me: “What’s the most important aspect of creating engaging content?”
Her: “Headline. If headline sucks, no one is going to read it.”
As a data professional, I know what people say doesn’t always reflect in the numbers.
So we built a kick-ass dataset of 100K long-form articles and headlines that got 100+ engagements.
We spent several weekends hand-picking writings of award-winning writers.
We categorized these writings.
Then came the fun part of ruthlessly removing all the biases from these articles.
Biases like seasonal and promotional articles.
We narrowed it down to 50K long-form articles that got 1000+ engagements.
Our award-winning data science team went berserk on this freshly minted dataset.
We used two of the best Machine Learning products in our analysis: Google and Watson.
Here’s what we uncovered:
Headlines, Personalization, & Engagement
We counted up the amount of personalized language in each headline. Personalized language is defined as the number of pronouns used in a text – number of pronouns that are either:
- Accusative (them)
- Genitive (my, mine, your)
- Nominative (I, you, he, she, it, they, and we)
In our last article about emotional tone and writing, I got a lot of emails with a request to offer a comparison between different categories of writings.
So we did exactly that. Here’s what we found:
This analysis is for headlines only (not body)
There’s a noticeable positive correlation between personalized language and engagement in the categories of Startup and Entrepreneurship.
Headlines, Active Tone of Writing, & Engagement
We calculated the score for an Active tone of writing by counting up the number of verbs used and dividing it by the total word count in the headline.
There’s a positive correlation between active language and engagement in the categories of Content Marketing (0.226), Finance (0.125), Startup (0.106), and Entrepreneurship (0.059).
None of the categories appear to have a noticeable negative correlation with active language.
Headlines, Quantitative Language, & Engagement
Next, we dived into exploring the relationship between engagement and usage of quantitative language in article headlines.
We calculated a quantitative/number score by checking whether or not an article headline contains a number in the form of a word or numeral.
Emotional Tone In the Headline:
We couldn’t end without looking at emotional tone – our bread and butter.
We combined all the attributes – Active Tone, Personalization, Quantitative.
We wanted to know which combination is the most effective combination for crafting an engaging headline?
And we got our golden answer.
For all the categories combined, we found that article headlines with good use of these three perform the best:
- Personalized language
- Active Tone of Writing
- Emotional tone of Conscientiousness
As per Watson documentation, “Conscientiousness is a person’s tendency to act in an organized or thoughtful way.” It constitutes of writing in a tone that conveys:
- Achievement striving / Driven
Let’s break this down by category:
This heat map reveals a couple of things about social media content:
- Conscientiousness has a negative correlation with social media
- Agreeableness has a negative correlation with social media
Data Science disclaimers:
- Correlation doesn’t imply causation.
- Use your own judgment when writing content.
- Don’t ask for our dataset. As our economist friends say, “The world’s most valuable resource is no longer oil, but data.
“Want to analyze how your writing stacks up on the emotional spectrum?