Optimization of Digital Content Engagement Patterns in Iran Using Genetic Algorithm and Survey-Based Data
Niloofar Mohseni
Sapienza University of Rome
1. Introduction
Grabbing and holding people’s attention in the today’s digital world, where they are continuously scrolling, tapping, and swiping through content, is more difficult than ever. What causes someone to halt and participate? What causes some posts to become viral while others receive very little attention? These are the questions that companies, marketers, and content producers face routinely.
Cultural leanings, platform popularity, and content type all have an impact on Iranian viewers’ digital commitment. However, we made the decision to adopt a data-driven strategy rather than depending on intuition or guesswork. We obtained actual data on content consumption patterns by conducting a countrywide poll with more than 5,000 answerers. We then calculated the optimal content strategy for maximum engagement using a genetic algorithm (GA).
The objective of this study was to be determining which platform and content kind work best to get the most views, comments, shares, and likes.
2. Knowing Iranians’ digital content consumption patterns
We required reliable data to work with before executing our algorithm. For better understanding how Iranians interact with digital material, their preferences, preferred platforms, and the kinds of information they like, we polled 5,034 of them. The questionnaire questions were designed to choose more than one option for some questions.
This is what we came on:
2.1. Who completed the survey?
Our study participants come from a mix of backgrounds:
Sex: 51% indicated they were female, 40% male and 9% did not identify.
Age: The largest cohort was aged between 18 and 25 (1670 responses), followed by those aged between 26 and 35 (1076).
Education: Most of participants had advanced degrees, with 2065 holding a bachelor’s and 2080 obtaining a master’s.
2.2. How much time do we spend online?
It should be no surprise that most people are drawn to their screens:
• 59% of them consume one to two hours of digital entertainment daily.
• 14% take 30 to 60 minutes.
• 23% of individuals spend over two hours per day.
2.3. What are the most popular platforms?
Most Iranians are on:
• Instagram – 3008 users
• Telegram – 2133 users
• YouTube – 1045 users
• Twitter – 834 users
• Aparat – 564 users
2.4. Content that people want to see
• News – 4835 people
• Entertainment (movies, humor, series) — 3926 people
• Educational – 3024 people
• Science & Technology – 2908 people
• Religious – 409 people
2.5. How do people interact with content?
• 41% of people share content often, while 33% never do.
• 40.5% of people are passive consumers of content—they nibble and scroll without engaging.
• 29.2% post to share, 12.6% make comments and 17.7% like.
• People want content that is: high graphic quality (41% ), short and simple (83%), captivating and attention-grabbing (21% ), educational and practical (40%), and humorous and entertaining (77%).
Now that we had a solid understanding of user behavior, it was time to feed this data into our genetic algorithm and let Python work its magic.
3. How our genetic algorithm (GA) works?
3.1. What is a Genetic Algorithm?
Genetic Algorithm (GA) is an optimization method based on the principle of natural selection. They emulates the process of natural selection of living beings across generations.
Good genotypes survive to the next generation and new individuals with beneficial traits are generated by GAs through crossover, recombination and mutation. Crossover combines segments from both parental chromosomes to generate variation, and mutation causes random alterations in an organism’s genes. GAs, can be applied in a more intentional manner through these principles, leading to greater impact in the digital space [1].
It functions as follows:
1. Content and platform pairings are made like a slot machine: “Entertainment +
YouTube,” “News + Instagram,” and so on.
2. A fitness function is used to test each one (as we’ll discuss in a moment).
3. The best combinations “reproduce” by combining components of multiple successful strategies.
4. There are some small tweaks that are made to spice it up.
5. This continues until the algorithm establishes its optimal strategy.
3.2. Measuring engagement: the Fitness Function
A fitness function determines the `good`ness of each content-platform combination. We tried to balance everything that matters to engagement, not just viewing. This was the formula we used:
F=(w1V+w2L+w3C+w4S)×Bcontent×Bplatform
Let’s break it down:
V = Views
L = Likes
C = Comments
S = Shares
w1,w2,w3,w4 = Weights controlling importance of each factor.
Bcontent and Bplatform = Bonuses due to survey data (to better represent true user experience).
Why this function?
• Engagement is prioritized (shares and comments trump passive views).
• It bundles by platform and type of content.
• It is flexible — its weights can be adjusted for different goals.
4. The results: what’s the best content strategy to drive engagement?
After using the GA on our dataset, we found: entertainment is the highest form of content. YouTube is the best platform.
Engagement Breakdown:
• View: 40.58%
• Like: 83.15%
• Comments: 77.65%
• Shares: 79.18%
Key Takeaways:
The most crucial part is the entertainment part. YouTube is the best format for growing engagement. The best stuff is short, funny and good to look at. The best content should be short, funny, and eye-catching.
4.1. Why YouTube scored highest
The algorithm selected YouTube because its weighted engagement score was greatest:
• High comment rate (77.65%) → YouTube is a platform for discussion in both human and textual (comments) forms.
• High share rate (79.18%) → Users share YouTube videos to other platforms often.
• Like-to-view ratio (83.15%, strong) → A substantial proportion of viewers engage with content by liking it.
• YouTube loves entertainment → Youtube has the best mechanics for videos, being one of the best platforms for both engaging entertainment content.
4.2. Why not Instagram or Telegram?
Despite Instagram having the most users surveyed, it did not have the highest overall engagement rating because:
• Less comments → Most people on Instagram will like content as opposed to
commenting on it with a low amount of users observing that comment.
• Lower shareability→ Stories are shareable, but Instagram doesn’t offer the same viral content spread potential as YouTube.
• Similar can be said for Telegram, which was (and still is) popular, but didn’t fare as well, because:
• Data leads to earthquakes )important and sensitive news(→ Telegram is not meant for entertainment, mostly news-related or educational content.
• Engagement per post → Users are consuming content passively, less interaction.
4.3. Conclusion
YouTube, the Comprehensive Data, a good mix of likes, Comments and Shares for Entertainment. Entertainment category videos have good engagement which leads to great user interactions. Instagram and Telegram are still really useful but might suit some other content types better (news, educational content, etc).
5.What this means & what’s next
So how can we make this information useful?
5.1. Guidelines for content creators
If you need more interaction, make fun YouTube videos. Short-form content, high quality, engaging, and brief works most effectively. Engage the audience with challenges, polls, and Q&A.
5.2. Future research ideas
What about the audience psychology? Future research could investigate why people pursue certain types of information. Intelligent out with artificial intelligence prediction of personalized engagement tactics that make it easier to guess which content people will be attracted to.
6. Final thoughts
As this study suggests, data is better than intuition. Rather than speculating what might work, we used poll data to identify the best content strategy for Iranian clients. YouTube is a platform that’s filled with entertainment content. But engagement patterns vary and more research like this should be done.
Reference
[1] Datacamp. (2024, July 29). Genetic algorithm: Complete guide with Pythonimplementation. Datacamp. https://www.datacamp.com/tutorial/genetic-algorithm-python