Anime Series Analysis Solution: Series Analysis (Per-Series Breakdown & Comparison)

お問い合わせ

【Extra】Anime Series Analysis Solution：Series Analysis (Per-Series Breakdown & Comparison)

We provide viewing data analysis not only per episode, but also for complete cour-based viewership (such as 3-episode early-phase clusters).

For late-night, adult-targeted anime, it was previously difficult to secure sufficient sample size, making it hard to compare between shows or detect changes across episodes. By utilizing REGZA’s viewing data, we have developed a package product containing a wide range of comparative indicators, based on over 4 million samples nationwide and about 670,000 samples in the Kanto region (aligned with TOKYO MX’s broadcast coverage), enabling comparison of programs aired on major networks.

Furthermore, by distinguishing between family and individual usage of television, we are also able to identify characteristics by gender and age group.

Key Features of the Package (focus: adult-oriented anime)

1. Comparability
Unlike streaming-exclusive shows, this data aligns major anime series side-by-side for apples-to-apples comparisons.

2. Precision Viewing Data
Instead of memory-based online surveys, this provides precise viewing data measured in seconds.

3. Multi-Source Viewing Modes
Real-time viewing, playback of recordings, and timer recording can all be compared and analyzed from a single data source.

4. Demographic Reception Analysis
Viewer acceptance can be compared by gender and age group (Teen, M1, M2, M3, F1, F2, F3*, and region-wide data).
*M = Male, F = Female, 1 = age 20 to 34, 2 = age 35 to 49, 3 = age 50 to 74

5. Episode-Level Trends
Fluctuations by episode (broadcast slot) and the exclusion of numerical drops caused by special programs or recap episodes are implemented, making it possible to compare the true strength of each series.

6. Large Sample Size in Kanto Region
Approximately 670,000 samples from the Kanto region (Tokyo, Kanagawa, and Saitama, including TOKYO MX coverage) enable not only tracking viewership trends per episode, but also analysis and comparison of viewers who watched a whole cour.
*Viewers who watched at least 70% of episodes (e.g., 9 out of 12 episodes) are counted.

7. Cross-Watching & Viewer Retention
By cross-analyzing groups of viewers who thoroughly watched certain series, it is possible to identify other shows that were heavily co-viewed, as well as track continued viewing on specific devices.

For late-night, adult-targeted anime, it was previously difficult to secure sufficient sample size, making it hard to compare between shows or detect changes across episodes. By utilizing REGZA’s viewing data, we have developed a package product containing a wide range of comparative indicators, based on over 4 million samples nationwide and about 670,000 samples in the Kanto region (aligned with TOKYO MX’s broadcast coverage), enabling comparison of programs aired on major networks.

Furthermore, by distinguishing between family and individual usage of television, we are also able to identify characteristics by gender and age group.

Key Features of the Package (focus: adult-oriented anime)

1. Comparability
Unlike streaming-exclusive shows, this data aligns major anime series side-by-side for apples-to-apples comparisons.

2. Precision Viewing Data
Instead of memory-based online surveys, this provides precise viewing data measured in seconds.

3. Multi-Source Viewing Modes
Real-time viewing, playback of recordings, and timer recording can all be compared and analyzed from a single data source.

4. Demographic Reception Analysis
Viewer acceptance can be compared by gender and age group (Teen, M1, M2, M3, F1, F2, F3*, and region-wide data).
*M = Male, F = Female, 1 = age 20 to 34, 2 = age 35 to 49, 3 = age 50 to 74

5. Episode-Level Trends
Fluctuations by episode (broadcast slot) and the exclusion of numerical drops caused by special programs or recap episodes are implemented, making it possible to compare the true strength of each series.

6. Large Sample Size in Kanto Region
Approximately 670,000 samples from the Kanto region (Tokyo, Kanagawa, and Saitama, including TOKYO MX coverage) enable not only tracking viewership trends per episode, but also analysis and comparison of viewers who watched a whole cour.
*Viewers who watched at least 70% of episodes (e.g., 9 out of 12 episodes) are counted.

7. Cross-Watching & Viewer Retention
By cross-analyzing groups of viewers who thoroughly watched certain series, it is possible to identify other shows that were heavily co-viewed, as well as track continued viewing on specific devices.

Screen 1：Overall Rankings Compared by Viewers Who Watched a Certain Number of Episodes

As shown by the orange bar graph, comparing the number of viewers who watched, for example, 9 or more out of 12 episodes allows us to assess the actual strength of the content (season).

In contrast, the green bars represent shows with many viewers who didn’t watch all episodes. These may have high ratings, but they don't accurately reflect the program's value as a whole content (season).

When the green bars are longer than the orange ones, it shows higher reach, often due to being broadcast on a key national network, which also means higher broadcasting costs. While this is good for recognition, for reaching a wider audience. But for purposes like overseas distribution and streaming promotion, the orange bar comparison is more meaningful.

As shown by the orange bar graph, comparing the number of viewers who watched, for example, 9 or more out of 12 episodes allows us to assess the actual strength of the content (season).

In contrast, the green bars represent shows with many viewers who didn’t watch all episodes. These may have high ratings, but they don't accurately reflect the program's value as a whole content (season).

When the green bars are longer than the orange ones, it shows higher reach, often due to being broadcast on a key national network, which also means higher broadcasting costs. While this is good for recognition, for reaching a wider audience. But for purposes like overseas distribution and streaming promotion, the orange bar comparison is more meaningful.

Screen 2：Gender & Age Balance – Demographic Strength by Series

By comparing gender and age group rankings for the four series shown in the graphs, a series labeled as 'Strong Across All Groups' ranked 2nd overall, while a 'Female-Dominant' series performed well with F1, F3, and Teen demographics, ranking 17th overall. A 'Strong in F1' series was popular among F1 but not at all in other segments, ending up 45th overall. Likewise, an 'Strong in M1' series was strong only among M1 and ranked 40th overall.

While it’s important to recognize series that perform well in specific demographics, these graphs also help identify which segments a series captures and how strength across multiple segments can elevate the overall ranking. The x-axis of the graph displays: Total Region, Total Family, M3, M2, M1, Teen, F1, F2, and F3, from left to right.

By comparing gender and age group rankings for the four series shown in the graphs, a series labeled as 'Strong Across All Groups' ranked 2nd overall, while a 'Female-Dominant' series performed well with F1, F3, and Teen demographics, ranking 17th overall. A 'Strong in F1' series was popular among F1 but not at all in other segments, ending up 45th overall. Likewise, an 'Strong in M1' series was strong only among M1 and ranked 40th overall.

While it’s important to recognize series that perform well in specific demographics, these graphs also help identify which segments a series captures and how strength across multiple segments can elevate the overall ranking. The x-axis of the graph displays: Total Region, Total Family, M3, M2, M1, Teen, F1, F2, and F3, from left to right.

Screen 3：Program Trends – Compare Episode-by-Episode Reservation, Playback, and Real-Time Viewing Against Overall Average

In the upper left graph of each program, you can compare episode-by-episode variations using a single-source dataset. The trends in timer recordings, total viewership, real-time viewing, and playback viewing are shown. An upward-sloping graph is rare - typically, viewership declines from episode 1.

The bottom left graph shows larger variations using episode 1 as the baseline (set at 100). In this particular case, timer recordings and playback viewership rise significantly. Faint colored lines below indicate overall averages. Since it is common for viewership to decline from episode 1, you can see how exceptional the upward trend is for this series.

The bottom right graph compares viewer engagement across demographics for real-time viewing, playback, and timer recordings. In this case, M3 and M2 show particularly strong engagement.

In the upper left graph of each program, you can compare episode-by-episode variations using a single-source dataset. The trends in timer recordings, total viewership, real-time viewing, and playback viewing are shown. An upward-sloping graph is rare - typically, viewership declines from episode 1.

The bottom left graph shows larger variations using episode 1 as the baseline (set at 100). In this particular case, timer recordings and playback viewership rise significantly. Faint colored lines below indicate overall averages. Since it is common for viewership to decline from episode 1, you can see how exceptional the upward trend is for this series.

The bottom right graph compares viewer engagement across demographics for real-time viewing, playback, and timer recordings. In this case, M3 and M2 show particularly strong engagement.

Screen 4：Fan Cluster Analysis – Understand Which Other Series Are Watched by Viewers of a Specific Title

For each program, graphs are generated based only on its core fan group (orange bar in Screen 1), showing how those viewers combine it with other titles. These charts require a large sample size to be reliable. The further to the right a title appears, the higher its total viewership; the higher it appears, the more disproportionately it was viewed by this fan group (indicating a strong lift).

In the example below, 'I'll Become a Villainess Who Goes Down in History' was watched 10 times more often by this fan group compared to the overall anime audience. Also, a cluster of “Isekai” (alternate world) titles is shown on the left, viewed 3 to 7 times more by this group than the general viewer base.

For each program, graphs are generated based only on its core fan group (orange bar in Screen 1), showing how those viewers combine it with other titles. These charts require a large sample size to be reliable. The further to the right a title appears, the higher its total viewership; the higher it appears, the more disproportionately it was viewed by this fan group (indicating a strong lift).

In the example below, 'I'll Become a Villainess Who Goes Down in History' was watched 10 times more often by this fan group compared to the overall anime audience. Also, a cluster of “Isekai” (alternate world) titles is shown on the left, viewed 3 to 7 times more by this group than the general viewer base.

Screen 5：Per-Series Summary (Overall Performance) – Playback Rate After Reservation, 1-Episode/3-Episode Drop-Off Avoidance, etc.

By measuring how much content is actually played back after timer recorded, we can estimate how engaging a series is. Additionally, the rates at which viewers avoid dropping the show after the 1st or 3rd episode can be compared against the seasonal average using both percentage rates and deviation scores (with 50 representing the average).

By measuring how much content is actually played back after timer recorded, we can estimate how engaging a series is. Additionally, the rates at which viewers avoid dropping the show after the 1st or 3rd episode can be compared against the seasonal average using both percentage rates and deviation scores (with 50 representing the average).

Package Conditions and Delivery Details

The graphs and data introduced here represent only a portion of the full package. The complete offering includes around 20 graphs and data visualizations (sample size: approx. 670,000), along with additional analytical tools.

＜Target Area＞
Tokyo, Kanagawa, and Saitama prefectures
For adult anime, the area is limited to these three to align with the TOKYO MX broadcast zone.*

＜Target Channels＞
8 terrestrial digital channels
(NHK General, NHK Educational, NTV, TBS, Fuji TV, TV Asahi, TV Tokyo, and TOKYO MX1)

＜Number of Series＞
Approximately 50–70 titles per cour
Varies by season. Children’s programs, long-running shows, or nationwide mainstream series are excluded.**

＜Delivery Timing＞
One delivery per cour, summarizing all episodes at season's end.
Example: For an April–June cour, delivery occurs around late July.
If the final episode airs in July, one additional week of playback data is included. Compilation begins on the 8th business day and requires about 3 days for validation. For instance, if the final episode airs on Sunday, July 6, data compilation would start July 14 (Monday), and delivery would be around July 17 (Thursday).

＜Delivery Format＞
・Multi-series pack (toggle between all titles) ⇒ Tableau file
・Single-series pack (fixed report per title) ⇒ PDF file

* Sample size may vary depending on the analysis period.
** Short-format programs may be excluded depending on the analysis period.
If purchasing the multi-series pack, Tableau Reader (free) must be downloaded and installed. A 64-bit compatible PC is required.
Publishing on Tableau Cloud is also possible, but additional licensing fees apply during the usage period.(For regular contracts, using the online version is convenient)

The graphs and data introduced here represent only a portion of the full package. The complete offering includes around 20 graphs and data visualizations (sample size: approx. 670,000), along with additional analytical tools.

＜Target Area＞
Tokyo, Kanagawa, and Saitama prefectures
For adult anime, the area is limited to these three to align with the TOKYO MX broadcast zone.*

＜Target Channels＞
8 terrestrial digital channels
(NHK General, NHK Educational, NTV, TBS, Fuji TV, TV Asahi, TV Tokyo, and TOKYO MX1)

＜Number of Series＞
Approximately 50–70 titles per cour
Varies by season. Children’s programs, long-running shows, or nationwide mainstream series are excluded.**

＜Delivery Timing＞
One delivery per cour, summarizing all episodes at season's end.
Example: For an April–June cour, delivery occurs around late July.
If the final episode airs in July, one additional week of playback data is included. Compilation begins on the 8th business day and requires about 3 days for validation. For instance, if the final episode airs on Sunday, July 6, data compilation would start July 14 (Monday), and delivery would be around July 17 (Thursday).

＜Delivery Format＞
・Multi-series pack (toggle between all titles) ⇒ Tableau file
・Single-series pack (fixed report per title) ⇒ PDF file

* Sample size may vary depending on the analysis period.
** Short-format programs may be excluded depending on the analysis period.
If purchasing the multi-series pack, Tableau Reader (free) must be downloaded and installed. A 64-bit compatible PC is required.
Publishing on Tableau Cloud is also possible, but additional licensing fees apply during the usage period.(For regular contracts, using the online version is convenient)