Data visualization reveals how Game of Thrones season 6 can halt falling ratings

Posted by: Yellowfin Team

By Eshan Wickrema and Lachlan James

While many Game of Thrones (GoT) fans around the world are salivating at the imminent return of the hit series adapted from George RR Martin’s fantasy novels, A song of Ice and Fire, it seems the enthusiasm for the show may be waning, according to data analysis.

However, using data visualization technology, we can also assess factors that may be directly linked to show ratings, therefore pinpointing how show creators David Benioff and D B Weiss can stop the apparent slide.

Game of Thrones rating per episode by season (according to IMDb)

Insights

Season five of GoT produced the equal highest rating episode of the series so far, with episode eight (‘Hardhome’) rated 9.9 by viewers on IMDb.
However, season five of GOT had a number of particularly low-scoring individual episodes:
- Season five produced the lowest rated episode of the entire series, with episode six (‘Unbowed, Unbent, Unbroken’) receiving a rating of 7.8.
- The season five finale was the lowest concluding episode of the saga to date.
  - Season 1: Episodes one, two, three, five, six, seven, nine and ten of season one outranked the corresponding episodes in season five
  - Season 2: Episodes one, three, five, six, nine and ten of season two outranked the corresponding episodes in season five
  - Season 3: Episodes one, three, four, five, six, nine and ten of season three outranked the corresponding episodes in season five
  - Season 4: Episodes one, two, three, six, seven and ten of season four outranked the corresponding episodes in season five

Season 5: A season of deviation 

Season five of GoT contained a huge amount of episode-by-episode fluctuation, producing both the equal highest rating episode of the series so far (episode eight, ‘Hardhome’, rated 9.9 on IMDb), as well as the lowest rated episode of the entire series (episode six, ‘Unbowed, Unbent, Unbroken’, rated 7.8 on IMDb).

The season five finale was the lowest scoring concluding episode of the saga to date, with IMDb raters most probably rebelling against the sudden death of fan favorite, Jon Snow. Season five also contained a string of comparatively lowly-rated episodes, including a particularly lean run at the start of the season. From episodes one through to six (inclusive), only episode two of season two (‘The Night Lands’) rated lower in an episode-by-episode comparison across the five seasons to date.

Those statistics suggest that HBO will be keen to start season six with a bang, while also avoiding such fluctuations and episode-by-episode deviations. So, how did the mixed episode-by-episode fortunes of season five impact its overall rating?

Game of Thrones average rating per season (according to IMDb)   

While the number of GOT faithful has steadily increased season-to-season, according to official HBO viewership figures and online download numbers from Torrent, its viewer ratings have stumbled.

Insights

The last season of GoT, season five, had the lowest average rating of any season yet (8.74), according to IMDb reviews
The season before that, season four, recorded the highest average rating on IMDb (9.23)
The first season of GoT recorded the third highest IMDb rating (9,00), and was the only other season to achieve an average rating of nine or above

Most curiously, unlike viewership, the overall season rating has not increased each season – season four had the highest average IMDb rating (9.23), followed by seasons one (9.00), three (8.94), two (8.86) and five (8.74).

So, it begs the question: What factor(s) in the Game of Thrones series have produced this ratings pattern? And, what clues (might) have show produces latched onto in order to arrest the notable ratings nosedive that occurred in season five?

Game of Thrones deviations from George RR Martin’s novels vs viewer ratings on IMDb per episode

Source: Viewer ratings data was taken from IMDb, while data regarding deviations between George RR Matin’s novels and the HBO GoT series was sourced from gameofthrones.wikia.com

Insights

Season five contained the fewest plot deviations and also received the lowest average viewer rating
However, whilst true for season five, the relationship between the number of plot deviations and average rating is not consistent from season to season
There are also examples of episodes with a high number of plot deviations that have received both comparatively high and low scores. This variance is also true for episodes with a low number of plot deviations.

Game of Thrones fans appear to assess deviations on their merits

At first glance, it seems that fans may have rebelled against the predictable nature of HBO’s season five of GoT, which contained the fewest number of deviations (122 total deviations, average of 12.2 per episode) from Martin’s series of novels, A Song of Ice and Fire. And, this might in fact be something for HBO to consider – season one scored the second highest average rating (9.00) and contained the second most plot deviations (236), with season two receiving the second lowest average rating (8.86) and containing the second fewest plot deviations (224).

However, this correlation isn’t consistent for season three (third highest average rating (8.94) and the highest number of plot deviations (352)) and season four (highest average rating (9.23) and third highest number of plot deviations (230)).

Deviations from novels:

Season 1: 236 total deviations, average of 23.6 per episode
Season 2: 224 total deviations, average of 22.4 per episode
Season 3: 352 total deviations, average of 35.2 per episode
Season 4: 230 total deviations, average of 23 per episode
Season 5: 122 total deviations, average of 12.2 per episode

Average rating per season (according to IMDb):

Season 1: Average per episode IMDb rating of 9.00
Season 2: Average per episode IMDb rating of 8.86
Season 3: Average per episode IMDb rating of 8.94
Season 4: Average per episode IMDb rating of 9.23
Season 5: Average per episode IMDb rating of 8.74

GoT episodes with a high number of deviations from the corresponding portions of Martin’s novels have received both comparatively high and low ratings by viewers on IMDb. For example, the first episode of season three, Valar Dohaeris, contained 43 differences from the book and was rated a relatively low 8.7 by viewers on IMDb. Conversely, episode nine in season three (The Rains of Castamere – also famous known as the ’Red Wedding’) contained 45 deviations from the book series, but was the equal highest rating episode to date on IMDb (9.9).

On the other end of the scale, HBO’s GoT series also contains examples where episodes have closely mirrored the events of the book series, receiving a wide range of relatively high and low IMDb ratings. The sixth episode of season five (Unbowed, Unbent, Unbroken) comprised just 12 deviations from Martin’s novels and received the lowest score of the entire series on IMDb (7.8). The fifth season also provides a prominent example of the opposite, with episode eight (Hardhome) incorporating the lowest number of changes of any episode to date (6), while simultaneously receiving the equal highest rating of any episode on IMDb (9.9).

So, if counting and comparing the number of deviations from Martin’s novels is an unreliable way to determine the underpinning factors that made GoT season five the most lowly rated yet, what is?

Major character deaths per season vs average viewer ratings on IMDb per season

Source: Data for major character deaths per season was obtained from the washingtonpost.com
Disclaimer: This data assumes that Stannis Baratheon and Jon Snow have actually been killed (there is much fan speculation and suspicion regarding the finality of these apparent ‘deaths’)

Insights

There seems to be a strong correlation between the number of main characters who die per season and the average IMDb viewer ratings per season
- Essentially, the more main characters killed each season, the higher viewers rated that season

Are we simply subconsciously baying for blood?

We’re all still in a state of bereavement and disbelief. Has Jon Snow (Kit Harington) and his luscious locks really deprived us of their presence on a permanent basis? If we take this at face value, and assume that Kit ain’t comin’ back, it makes for some interesting data analysis.

It appears as though, despite our apparent grief, we actually feed off the shocking demise of seemingly central characters. That is, the higher the main character head count per season, the higher the average rating per season.

For example, season one started with four major character deaths and enjoyed an average rating of nine. Season two only killed-off one main character (Renly Baratheon) and its rating dipped to 8.86. Season three upped-the-ante with five significant players meeting their demise, which saw the average rating increase to 8.94 for that season. Season four took it up another notch, disposing of seven central characters (most of any season), which saw the average season rating jump to its highest recorded point (9.23). Season five reduced major character fatalities to three, which saw a simultaneous dip in viewer ratings (8.74) – the lowest ever average for any season so far.

So what does this apparent trend say about us? Do we thrive on the controversy of character upheaval? Or are we simply a bunch of sadists at heart?

Tweet us your thoughts @YellowfinBI

We recommend

Data visualization assesses kids’ enthusiasm for Christmas >