The basic idea of viral content is simple: create original content that adds value, making people want to share it. That’s a lot easier said than done. I have analyzed the 60 best articles from medium.com, using a script that automatically extracts the information from the articles, looking for the common thread between them.
First, I want to delve into the most important part of the text extraction program: the extraction and processing of text. To obtain the information, the software “reads” the articles simply by looking at the structure of the core HTML code. Using this methodology, all of the medium articles followed the same pattern and were easy to analyze.
For the extraction of useful information, I used PYTHON NLTK, a solution specifically designed for processing written language. Through this analysis, I was able to find commonalities between all of these articles. Below are a few of the biggest takeaways from the analysis of how to create viral content.
Creating Viral Content
You should be able to read it in 4 minutes
The length of the content is very important. Nowadays, people will not read anything that takes longer than four minutes. The articles that we analyzed had a maximum length of 2360 words. The software counted the words per article. Below is a graphic visualization.
The majority of the articles were between 400 and 1000 words.
It’s good to have pictures but don’t abuse them
Take a look at the 11 articles without any pictures but have generated huge engagement with the public. You don’t always need pictures! However, the best practices are to have between 1 and 5 images or graphics to connect with your reader.
The outlier with respect to images is this one that has 44. In my humble opinion, this was successful because scrolling through images is very quick and entertaining.
The articles on medium.com use the HTML technique called figure for elements that are not text. This created a lot of work for our text analysis bot but it was eventually successful.
Talk about topical issues
The software we developed extracted tags in the text from each article to determine what they were talking about. The conclusion was that medium is a meeting place for entrepreneurs in general. Articles on the site talk about technology, design and bettering yourself.
We have also collected data on a few topics that were infrequently used in these articles. This is because they are “niche topics” and exist to attract readers who are looking to learn. A few of them are Metrics, Education, Parenting, Politics.
Tell a story and create protagonists
At the same time, I have analyzed the words that repeated in each article. Almost all of the articles had one common denominator: people. Our bot showed that “I” and “you” were within the top 50 words used in each article. They do this because it’s important to make sure the audience knows that they can benefit from what they’re reading.
Apart from I and you the words startup, tech, design, people, and entrepreneurs were used in almost all of the articles.
The most common words in almost all the articles are “I” and “you”. Tell a story with real characters.
What more can you do thanks to this type of software?
Text analysis allows us to innovate on many everyday tasks, since it is possible to make a computer/smartphone understand text and render it as if it were a human. In Secmotic we like to bet on new technologies like this and we have used it for various jobs in this area:
- Classifying text
- Analysis of feeling
- Analysis of text structure
This has been all about analyzing viral content. Do not miss the following post on datascience and chatbots!
We’d also love to answer any questions you have about the article or viral content!