Introduction
For creation platform such as YouTube and Twitch, it is particularly important to analyze the channel performance though data analysis. With the exploration of data, we can discover the overall performance of channel, competitive environment and future development direction and so on. The visualized data display form makes analysis more intuitive, more structured, and easier to communicate. More importantly, deeper problems can be found in the visualization results. Meanwhile, for content producers, all kinds of channels has already turned into the stage of highly competition, creating differentiated high-quality content videos has become one of the core survival methods.
Game design belongs to the professional content of game field, it has a high threshold for content creation. Continuous content output in this field can effectively contribute to establish industry influence or personal brand. Meanwhile, more and more people are learning game design, positions like game planner and game UI designer have become emerging occupation. Therefore, more high-quality content is needed for students and practitioners to learn, thus there is a broad prospect for this sub-division of markets.
This report compares and analyzes the dataset from five YouTube channels in game design field by data visualization, mainly focusing on video data and channel performance, so as to extract valuable strategies for content creators. As a note, the game design discussed in this report is a comprehensive game design category which mainly includes design, planning and development.
Inspiration
Considering the overall analysis of this visualization is based on the comparison among five channels, the research should focus on different metrics of channel performance respectively. Thus the visualization result will not be a single graph with large scale and complex internal structure, but a integrity composed of multiple small parts. On this basis, I looked at several visualization projects related with YouTube data analysis on Tableau Public in order to inspiration.
First reference came from YouTube- What people like on YouTube? (Sharma, 2020). This visualization focus on audience’s overall likes and dislikes for categories and videos of YouTube. The most interesting part is Most Loved Category by States, it colors each state’s degree of likes for video categories, highlighting the top five states which have enthusiasm for music and entertainment videos.
Second example used for reference is YouTube-Viral-Trends-COVID-19 (Suresh, 2020). It combines tables, bar chart, word cloud and maps, representing the data from various aspects of YouTube platform during the Covid-19. Through the usage of date filter, it allows us to interact with visualization directly.
Materials
The data from five YouTube channels was collected by YouTube Data Tools. The commonly used module of site is Video List module. This module helps users to create a list of videos information of a specific channel or a public playlist, which provides statistic data conveniently and rapidly. Google sheet was used to clean up and organize datasets from five channels.
Tableau Public is a widely used free visualization software. I used this tool for the visualization work of report. It should be noted that the datasets from five channels need to be combined for analysis by Tableau. Meanwhile, I also used Wordcloud, an online word cloud generator. Because I found that Tableau Public doesn’t have Pivot function as Desktop version, which limits the possibility of converting multiple word columns split up from columns to row. Therefore I used Wordcloud to visualize the frequently used words in video titles.
Analysis
Analysis Process
Choosing appropriate channels as research objects need to be carefully considered. Through the overview of search result of “game design”, I noticed that the videos with high views concentrated on several top channels, and these channels have top ranking subscribers in the meantime. In order to explore the successful experience for game design channel, I selected five representative channels from the channels with the largest number of view count. Selection criteria mainly includes: game design related channel; organization instead of individual, more than 100k subscribers.
The datasets created for analysis consists of both channels and videos information. Two datasets were combined together through the same channel titles. The main metrics of videos consists of published time, category, duration, view count, like count, dislike count and comment count.
In the process of visualization analysis, I planned to conduct comparative study of five channels from three perspectives, including overall activities, channel performance and content analysis, and then summarize key findings and put forward corresponding recommendation.
When launching project to Tableau Public, I realized that Tableau Public supports direct interaction between viewers and data panel. However, I had limited the display form within static graphs, instead of interactive data visualization. Therefore, I added filters on each pages of story, including channel title, published year, data range, etc. Thus viewers can directly operate and interact with data, filtering out the information which they are interested in.
In terms of visual design, I updated channel colors in iteration process. The first draft used default white background of system, which was relatively ordinary and plain, failing to convey the theme of game design. The second draft changed the background color to black and fine-tuned the channel colors, in order to create an atmosphere of game design.
Results
1. Subscribers and Interaction
The bar chart shows the more subscribers the channel has, the more view count and like count from viewers. The video count has no obvious correlation with interactions. GDC has the largest number of videos compared with other channels.
However, from data of per video, it can be seen that per video of Game Maker’s Toolkit has highest views and interactions, although it only has 147 videos. This shows that Game Maker’s Toolkit follows a strategy with low quantity but high quality. Meanwhile, GDC has as many as 1838 videos, which scattered the number of views to some extent.
Findings
The comparison between GDC and Game Maker’s Toolkit proves that it has limited effect when only having numerical superiority. The quality of video is more important than quantity, but both of them are indispensable.
2. View Count by Duration
This scatter diagram shows the views of videos from five channels by duration. It informs the viewer that the high viewed videos mainly concentrate on 400~1400sec which is 6min~23min. Significantly, Brackeys and Game Maker’s Toolkit have more high viewed videos than other channels.
It’s worth noticing that GDC has different shape of diagram compared with other channels. The duration of its high viewed videos didn’t concentrate within 400~1400sec, which means duration has little effect on its view count. This is principally because GDC (Game Developers Conference) is a conference of game field, its YouTube channel has massive amount of professional videos which form a knowledge base, differing form common YouTube channels. Its video views proportion is relatively even, this is likely because the audiences reach the video page through categories of channel homepage or search function of homepage.
Findings
Video’s duration between 6mins to 23mins usually attracts more views. Proper duration of video is a favorable factor for popularity.
3. Video Count and Frequency of Posts
This chart combined view count with video count. It shows that the view count varies with the quantity of videos roughly, but different channels varies.
Overall, five channels’ development started from 2015. The peak of views happened at 2017 (Brackeys and Game Maker’s Toolkit). The view count of Brackeys and Game Maker’s Toolkit both stayed at a high level. Although GDC kept the continuity of posts, its view count growth rate was still negative. After 2017, the view count of most channels decreased, which needs more fact to explain.
According to frequency of posts from each channel, it can be concluded that too few or too many posts every month do not help much in increasing views. Enlighted by the chart from previous page, apart from quality variables, 3~8 posts every month would be a good frequency.
Findings
The view count has no direct connection with video quantity. Based on guaranteeing high quality of videos, it helps a lot to keep the continuity of posts and post 3~8 videos every month.
4. Frequently Used Words of Video Titles
The word cloud of all video titles shows Design, Games, Flash, Unity, Art, etc. are frequently used words in titles. These words describe the main content of videos, focusing on game design, develop software and popular games.
In the meantime, I was also thinking the way to further explore the information in wordcloud. By filtering out the top 100 videos with highest views and turning their titles into wordcloud, I further discovered frequently used words in titles of most popular videos.
Most words in this wordcloud are the same as All Video Titles wordcloud, but there exists some outstanding features, such as the lack of “Flash”, more words emphasizing production like make and maker, games like Mario, GMPK and Zelda stand out.
Findings
Wordcloud inspired us the popular themes and hot topics of game design videos, which can be used as reference for YouTube content producers.
5. Most Popular and Most Engaging videos
The most viewed videos cover mainly how-to videos, tutorials and design evaluation about game design. Three of five top videos are from Barckeys. These three videos belong to same “How to make Video Game” series which provide practical experience for game designers and developers to learn.
The most engaging videos mainly concentrate on videos about game comparison, how to, rankings and bad game design. Some of videos have long lasting popularity among game design learners, such as How Game Designers Protect Players From Themselves (form Game Maker’s Toolkit) and Bloodborne vs. Dark Soul 3 (from Snoman Gaming)
Findings
Popular videos offer the audience with practical and useful content. Also videos with distinct theme and problem-solving videos are more popular. Specialized and differentiating content is the hinge to obtain core competition and achieve the competitive advantage and specialized channel.
UX Research
For user experience research, I conducted two remote interviews to figure out the usability of visualization interface and whether the project logic is clear enough. The objectives also included getting recommendation on the information conveyed and visual aesthetics from participates. I recruited two participates, one with rich gaming experience and one with experience on both vlog and game, both of them are graduate students.
Interviews were organized remotely and recorded by video recording. Participated were asked to view the visualization dashboard story and interact with interface on Tableau Public accompanied by the presentation of project content by me. During the process of interaction, participates were encouraged to speak out what they are thinking and questions they met. Post-test questions were asked to get advices about both visualization interface and report findings.
Findings and recommendations
1. “I think a dynamic display of these multiple charts would be easier to understand”
One participate thought a dynamic demonstration of key findings in visualization would be nice. The colorful bar charts can easily distract viewers, and users need to focus on key point or data by themselves, which increasing the difficulty of users to understand the visualization. Meanwhile, new users for Tableau Public haven’t used the filter function before, it would be better to give them automatic guides.
2. “I expect to know more information for the last few months or weeks”
One participate mentioned that he often looks for what he is interested in as the topic of video in the current hot events. Thus he expected this visualization research could present the ongoing trend or hot events. For example, popular games or popular topics in game field, making him identify the points he can explore or dig deep into among the hot topics.
3. “Hope this project can offer more recommendations for individual creator”
The same participate talked his experience on vlog. As an individual vlogger, it’s relatively challenging to make more than four high quality videos in part time within a month. If the video is solely game playback or discussion of new games, it can be done quickly. Therefore, it would be more helpful if this research can provide some suggestions for individual content creators.
Reflection
The visualization results showed successful experience and valuable strategies of top YouTube channels of game design. For future research direction, inspired by feedback from participates in UX research, I plan to focus on individual content creators, researching their creation situation on YouTube or other platforms, and taking in-depth study on the ecology of content platform. By visualization graphs, problems of creators and platforms should both be found, so as to help creators improve impact and benefits，and give advice for platform, helping to build a good creation environment in the same time.
This series of visualization research made me further recognized that data is not equal to information, only the processed data can turn into information, and the information presented by visualization is more in line with people’s habits of abstract thinking and image thinking. More importantly, visualization is to help users locate and find problems, it is not the same as stacking of charts. Only the graphs that convey the real significance can be regarded as visualization.
Tableau File Link
Reference
Sharma, V., 2020. Tableau Public. [online] Public.tableau.com. Available at: <https://public.tableau.com/profile/vishal.sharma1976#!/vizhome/YouTube-WhatpeoplelikeonYouTube/Dashboard1> [Accessed 10 December 2020].
Suresh, S., 2020. Tableau Public. [online] Public.tableau.com. Available at: <https://public.tableau.com/profile/siddharth.suresh8869#!/vizhome/YouTube-Viral-Trends-COVID-19/YT> [Accessed 10 December 2020].
Tools.digitalmethods.net. 2020. Youtube Data Tools. [online] Available at: <https://tools.digitalmethods.net/netvizz/youtube/> [Accessed 14 December 2020].