Recently I have been studying if I can predict the performance of marketing documents using a sentiment analyzer. I’m checking whether the emotional tone of a fintech marketing document affects how much engagement it gets on various platforms. I’m defining engagement here as likes, comments, and other types of reactions to posts.
In simple terms, this means I’m studying whether happier posts get more likes. My first experiments were done on LinkedIn posts but I plan to test this theory on other platforms later, including the blogging platforms Medium and Substack and even the fintechs’ websites themselves.
The Relationship Between Sentiment and Engagement
I’m using the Python library TextBlob to get the sentiment scores for posts and then using OpenOffice’s LINEST to create a regression formula for each influencer. In several of my experiments I got an R squared value between 0.50 and 0.60 when I was using sentiment polarity and subjectivity scores to predict engagement on individual fintech influencers’ LinkedIn posts. High subjectivity scores partially predicted the performance of an influencer’s post in comparison with their other posts.
I was comparing posts from individual fintech influencers in these experiments, so total engagement and reach still depended on other factors such as the size of the influencer’s network. The results were less accurate when I ran experiments that combined posts from multiple influencers. Sentiment polarity made each individual prediction formula more accurate, but the polarity factor was different for different influencers. The effect of high subjectivity was more consistent across influencers.
It’s unclear why higher subjectivity would give a LinkedIn post more reach. That metric indicates that an influencer’s talking about their personal feelings and opinions rather than stating facts. And LinkedIn is a business platform, not a social network designed for entertainment. But after thinking about it some more I think I figured it out.
The Importance of Storytelling
High subjectivity scores are one indication that the influencer’s telling personal stories. For example, they might be describing milestones in the history of their startup, discussing the challenges of launching a business, or providing case studies where they explain how their startup helped a prospective customer. These stories are more compelling than marketing posts that list the features that their fintech apps have, or sales posts that encourage readers to rush out and sign up for the service. They give readers a reason to learn more information about the company. And they help differentiate companies that appear very similar.
For example, payment processors, money transfer apps, and neobanks often have similar feature sets. These startups do offer features that traditional banks and financial service companies don’t offer, but their peers often offer the same features as well. For example, any money transfer app offers better exchange rates than banks do. So money transfer apps need another way to distinguish themselves, and one of them is by having their employees tell stories about the customers they’ve helped.
Money transfer apps are especially well-positioned for telling stories because of the services they provide. They help low-paid farm and restaurant workers send money back home. They can help wealthier people pay for airfare and hotel rooms in exotic destinations. They help expats set up normal lives when they’re working overseas and their clients are located around the world. And they even help businesses ship products to customers in foreign countries and collect payments across international borders.
And it’s well known that personal stories about business frequently go viral on LinkedIn. There are even marketers who are now mocking the personal stories told by other influencers. You might have seen some of these viral posts in your feed already. Stories about helping people find jobs or succeed in their interviews are especially common. Some of these LinkedIn stories are so popular that unethical posters copy them from other people and reshare them as their own content.
Further Research Goals
This article is meant to be an update on a larger project I’m working on, a short ebook about sentiment analysis in fintech marketing. I’m going to post that on my blog later once it’s complete, but I plan to study sentiment on other platforms before I do that. For the time being, though, I’ve learned that the sentiment analyzer has found a relationship between subjectivity and engagement and it appears that this is because personal stories contain more subjective terms than other posts.