Machine Learning in WordPress: How to Analyze the Emotional Impact of Your Content Using AI

With the right tools, you can actually quantify the emotional impact of your content and tweak it to target different feelings. To be fair, this isn’t an exact science, and we can’t tell you which emotions will resonate best with your audience. Still, you can learn a lot from the process.

In this article, we’re going to talk about how machine learning in WordPress may change the way you approach content creation. Then, we’ll introduce you to a tool that will help you analyze the emotional impact of your articles. Let’s get in touch with our feelings!

How artificial intelligence may change the way you use WordPress

First, it’s important to understand that we’re still nowhere close to real Artificial Intelligence (AI). While the technology has advanced, it’s still not able to write original content. However, one area in which we have made huge strides is machine learning, which is a (kind of) similar field.

Machine learning involves having computers use algorithms to get better at particular tasks. The more data you feed these systems, the better they can become at analyzing it and predicting results.

There are already a few WordPress applications available to the public that are based on machine learning. For example, there’s WordLift, which enables you to build better taxonomies and find sources for your content.

Most machine learning applications are still very hit and miss, however.

That means it’s hard to predict how the technology will progress in the short term, and how it will impact content creation.

To give you a better idea of where it’s at, we spoke to Tom Ewer – the founder of WordCandy – who has plenty of experience creating content for WordPress websites. Here’s what he had to say:

Machine learning in WordPress: An introduction to Watsonfinds

Watsonfinds

Author(s): Watsonfinds

Current Version: 2.0.0

Last Updated: May 8, 2019

watsonfinds.zip

100%Ratings 50+Installs WP 4.0+Requires

Watsonfinds is a service built on top of the IBM Watson system. Watson bills itself as a kind of AI, but to be more accurate it’s a sophisticated machine learning system with access to several terabytes of data, featuring the ability to analyze that data for specific purposes.

In this particular case, Watsonfinds can use its ‘knowledge’ to analyze any text in the WordPress editor and break it down into five emotions: joy, sadness, anger, disgust, and fear. Armed with that information, you can tweak your content until you’re targeting the exact sentiment you want (as long as it’s one or more of those five).

To use this tool, all you have to do is install and activate the plugin. Then go to any of your pages or posts, and open them using the Edit option. Inside, you’ll find a new button at the top of your editor. Click on it when you’re ready to analyze your content’s emotional impact:

Now, Watsonfinds will display a window including a simple analysis of your content. This includes a percentage breakdown of each emotion’s ‘strength’, and a brief tidbit about the ways your tone can affect engagement:

At this stage, you can go back to work on your content, and put your new tool for machine learning in WordPress into action!

Watsonfinds is a free plugin, but there’s a premium service in the works that will include more in-depth analysis.

How to analyze the emotional impact of your WordPress content using Watsonfinds

As we mentioned earlier, Watsonfinds focuses its analyses on five core emotions. In this section, we’re going to talk about each of them, and explain how you can use machine learning in WordPress to better connect with your users.

1. Sadness ☹️

No one wants to be sad, but it’s fair to say that in some cases, you may want to elicit strong negative emotions from your audience. The goal is not to play with their feelings, but to form an emotional bond with them. To illustrate that point, here’s a short paragraph from one of our recent articles that Watsonfinds deemed ‘sad’:

It’s easy to spot why Watsonfinds came to this conclusion. In fact, we’d go as far as to say that paragraph is outright depressing. In this case, the author used the text to form a connection with the reader, by discussing a problem we’ve all gone through. That way, when we eventually reveal that there’s a solution, the emotional impact will be more significant.

On the other hand, if you’re writing an article and you just jump right into the reveal, it’s often not as engaging. Notice our use of phrases such as “drains our motivation” and “lack of growth”. If you take those out, the sadness score decreases sharply:

However, Watsonfinds still manages to understand that the first part of the paragraph isn’t positive. That’s impressive when it comes to machine learning in WordPress.

2. Joy 😄

You’d be hard-pressed to find someone who doesn’t enjoy feeling happy, which makes this emotion an easy sell in your content. Let’s see how Watsonfinds fares, with a small excerpt from our recent article on themes that work great with site builders:

The positive spin in these two paragraphs is easy to spot. To give you more context, that excerpt is what we call a reveal. It’s a simple technique that we sometimes use in introductions, where we break down a problem, and then present the reader with its solution. Naturally, the reveal should be a joyous occasion, and Watsonfinds didn’t fail to notice that.

After having played around with machine learning in WordPress, we’ve found that a score of over 60% is pretty strong for any given emotion. It means that feeling is clearly predominant, but not so much that you’re beating readers over the head with it. Overall, Watsonfinds is especially adept at identifying ‘happy’ paragraphs, which is great if you want to make sure your content is hitting the mark.

3. Disgust 🤮

So far, we’ve talked about two rather straightforward emotions. Now it’s time to look into a more complex one – disgust. From a storytelling perspective, disgust doesn’t have as many applications as joy or sadness, but it can still come in handy. Here’s an example of a Salman Rushdie quote that Watsonfinds found predominantly ‘disgusting’:

Watsonfinds explains that this kind of emotion can be a useful tool when it comes to changing readers’ minds. However, in our experience, it’s far better to use disgust as a way to bond over things you don’t like. For example, if we were to review a WordPress plugin you hated, we might say something like:

We’d never be so blunt, of course, but this over-the-top example shows that machine learning in WordPress does struggle a bit when it comes to analyzing disgust. During our tests, we had to resort to using extreme adjectives to get our disgust score up, which isn’t ideal.

4. Fear 😨

To keep our trend of negative emotions going, let’s talk about fear. It’s not something you want to overuse, but it can be a handy tool when use intelligently in writing and marketing (FOMO, anyone?). Here’s an example of an excerpt from a recent article about secure WordPress themes:

We talked previously about how you can use sadness to make a stronger impression in your introduction, but fear also does the trick. In this case, the author mentioned the things that could go wrong if you didn’t use a safe theme, and then they swept in and told you how to deal with it.

The key when it comes to eliciting fear is talking about adverse outcomes and how they can negatively affect your readers. If you can tap into the things they’re scared off, you’ll have an easier time engaging them with solutions.

5. Anger 😠

Last but not least, let’s talk about anger. To be honest, anger is rarely one of the emotions we try to elicit from readers when writing articles. Anger is a tricky tool, which can help you connect with your audience but can also alienate them. Let’s create a short example of a WordPress theme review, and see what Watsonfinds has to say:

According to Watsonfinds, that only scores a 55.34% in the anger category, which is surprising because it reads like an enraged YouTube comment. The tool also discusses how anger can prompt readers to take action, which may or may not be the case.

In our experience, if you’re angry about the same things as your audience, you can use that to form a connection, but it’s not a sustainable emotion. As far as your content is concerned, emotions like sadness, fear, and joy are easier to quantify and to get right when using machine learning in WordPress.

Conclusion

If you can make your readers feel something, you can engage with them and build trust more easily. Watsonfinds helps you analyze the emotional impact of your content, and as far as we’re concerned, it does a decent job. It doesn’t tell you what to change, but it does signal if you’re targeting one of the core emotions correctly.

Let’s break down how each of those five emotions can affect your content:

1. Sadness: Sad feelings can make your audience feel closer to you.
2. Joy: Happy readers are the best kind of audience, and Watsonfinds is excellent at identifying positive content.
3. Disgust: This emotion is a bit tricky, but can help you bond with readers over shared distastes.
4. Fear: Despite its negative connotations, fear is a highly effective motivator for driving conversions.
5. Anger: This emotion can, in some cases, help you bond with readers (although we think it’s best to avoid it in most cases).

What do you think about the whole idea of incorporating machine learning in WordPress and your content creation processes? Have you heard of any other interesting implementations of the idea?

Don’t forget to join our crash course on speeding up your WordPress site. Learn more below:

Layout and presentation by Karol K.