Rooster Teeth Podcast, cast analysis

Analyzing the impact on Views, Likes, and Dislikes from different cast members on the RT podcast

Image source: roosterteeth.com/show/rt-podcast

Overview:

Main Purpose: To analyze the impact on Views, Likes, and Dislikes (VLD) from different cast members on the RT podcast

Additional Purposes: 

1. Analyzing the affect tags have on VLD
2. Developing different systems to analyze impact performance from VLD

Means: The RT video podcast is compiled neatly by the Rooster Teeth Video Playlist. Additionally, cast members are labeled with their name found in the tags section. Using Social Blade, python HTML scrubbing, and a MATLAB database structure, each video on the playlist had their VLD data and tags downloaded. Tags were counted as entries into the database. If it existed, it would accumulate the VLD data associated to the tag, while if it did not exist, it would create a new entry with the VLD data. 

Disclosure: This analysis does not include data from podcast #241-332. During this time Rooster Teeth did not tag their cast members, and would require manual data entry.

Like many other YouTube shows, the RT podcast is reliant heavily on the personalities and talent that rotate throughout. While the podcast holds true a 4 person team for each episode, who consists of that team varies. The classic "A-team" of Gus Sorola, Barbara Dunkelman, Burnie Burns, and Gavin Free is the most frequent and expected, however many different guests and "B-team" members have been brought on. While this A-team is well known and indisputable, the data backs up the obvious claim as well, With the A-team taking up all 4 spots on the highest number of appearances (1).  The next highest number of appearances is seconded by Jack Pattillo, with only 30 appearances, a far second from Barbara Dunkelman's 82 (1).

"A-team" appearances

The A-team also accounts for 442 out of the 667 logged appearances, a 66% domination 









Due to the high number of appearances, it makes sense as to why the A-team also rank the in the top 4 for views, likes, and dislikes. This is where different "scoring systems" come in to play. Ratios between appearances, views, likes, and dislikes (AVLD) even out the playing field for all cast members and shows projected data. For example, a View/Appearance outlook would show the projected amount of views a podcast would receive if a certain cast member would appear. Another reason why ratio based scoring systems are important are because of flat line data and accumulated data. For example, earlier podcasts have little to no fluctuation to their AVLD stats however newer podcasts do. As long as the assumption that audience members over time reflect the same data, ratio based scoring systems provide a much more accurate picture. While a full analysis of every cast member is too long for this post, the "best" cast selection and "poorest" cast selection will be analyzed, looking at the top members in each category, as well as their consistency. 

Likes/Dislikes

Here, the Likes per Dislike shows how well the audience receives a cast member. A high L/D score means that they received a high number of likes with very little dislikes. The higher the score, the more well received, vice versa. Here is the graph reflecting L/D scores as well as the number of appearances. The number of appearances is crucial, because it is the analysis of accumulated data. A single data point is not sufficient to make an assumption. A high L/D is preferable. 

The A-team sticks out here in terms of appearances, however ranks around midfield for audience reception, averaging 36.65 likes per dislike. 

Best performing cast in terms of L/D: 

Griffon Ramsey: 6 appearances, 69.5 L/D

Geoff Ramsey: 7 appearances, 60.6 L/D

Jordan Cwierz: 5 appearances, 54.9 L/D

Jack Pattillo: 30 appearances, 52.6 L/D

Poorest performing cast in terms of L/D:

Aaron Marquis: 3 appearances, 11.5 L/D

Brandon Farmahini: 22 appearances, 14.3 L/D

Jon Risinger: 10 appearances, 16.2 L/D

Chris Demarais: 26 appearances, 17.3 L/D

Views/Appearance 

Appearance based ratios are trickier because of accumulated data. Because older podcasts hold higher VLD data simply because of greater exposure time, those who were featured on older podcasts will be favored in terms in of VLD data. Views per appearance is one of the, if not the most valuable stat to have. Views generate revenue. A high V/A is preferable

Best performing cast in terms of V/A: 

Griffon Ramsey: 6 appearances, 875,343.3 views/appearance
Geoff Ramsey: 7 appearances, 806,700.5 views/appearance 
Michael Jones: 12 appearances, 714,257.7 views/appearance
Joel Heyman: 10 appearances, 646,126.7 views/appearance

Poorest performing cast in terms of V/A:

Adam Kovic: 2 appearances, 339,020.5 views/appearance 

Aaron Marquis: 3 appearances, 445,087.3 views/appearance 

Jon Risinger: 10 appearances, 462,139 views/appearance 

Brandon Farmahini: 22 appearances, 475,682.6 views/appearance

Likes/Appearance 

Likes per Appearance is again going to favor older podcasts that have had more exposure time, however is included. A high like per appearance ratio shows active liking for a cast member and passive disliking for a cast member. A high L/A ratio is preferable. 

Best performing cast in terms of L/A: 

Michael Jones: 12 appearances, 10,817.5 L/A

Jordan Cwierz: 5 appearances, 9923.8 L/A

Griffon Ramsey: 6 appearances, 9876 L/A

Jack Patillo: 30 appearances, 9713.9 L/A

Poorest performing cast in terms of L/A:

Aaron Marquis: 3 appearances, 5545.3 L/A

Jon Risinger: 10 appearances, 6519.3 L/A

Brandon Farmahini: 22 appearances, 6794.7 L/A

Sally Le Page: 4 appearances, 6901.25 L/A

Dislikes per Appearance

High dislikes per appearance ratios show active disliking of a cast member and low ratios show passive like for a cast member. A low D/A ratio is preferable. 

Best performing cast in terms of D/A: 

Griffon Ramsey: 6 appearances, 142.2 D/A

Geoff Ramsey: 7 appearances, 155.7 D/A

Matt Hullum: 4 appearances, 161.25 D/A

Jordan Cwierz: 5 appearances, 180.8 D/A

Poorest performing cast in terms of D/A:

Aaron Marquis: 3 appearances, 482.33 D/A

Brandon Farmahini: 22 appearances, 474.63 D/A

Chris Demarais: 26 appearances, 406 D/A

Jon Risinger: 10 appearances, 402.3 D/A

Flaws

Data flat lining and exposure time does skew data as discussed. Audience bases are also a big issue that should be mentioned. This is an edit of the original post, as some of the forecasting data might not be as applicable. Due to how long the RT podcast has been running, the audience of today is not like it was in the previous years. After further analysis, a trend of a downward audience reception was noted. In further pursuit, another program was thrown together to gather purely statistical information. It appears that audience reception is down. 

Noted here, there is an obvious low spot for the past 50 episodes, as well as increased spikes bias. This could be linked to audience rejection of "b-team" based cast and an acceptance of "a-team" cast. Some shift occurred to where the RT podcast had a polarized audience. 

Conclusion

While this data can give a big insight on how cast members affect the RT podcast, there are many other factors that are unaccounted for. Conversation topics, special events, timelines, sponsors, etc. can all affect the performance of a podcast. 

Sources/Extra:

-Here is the list of all cast members analyzed, as well as other stats available

-Because the information was stored in tags, the python script also gathered 1575 unique tags that can also be analyzed

- Here is also a more recent list analyzing from episode 350 until now.