What Players Want: Information Needs of Players on Post-Game Visualizations

Abstract

With the rise of competitive online gaming and esports, players’ ability to review, reflect upon, and improve their in-game performance
has become important. Post-play visualizations are key for such improvements. Despite the increased interest in visualizations of
gameplay, research specifically informing the design of player-centric visualizations is currently limited. As with all visualizations,
their design should, however, be guided by a thorough understanding of the goals to be achieved and which information is important
and why.

This paper reports on a mixed-methods study exploring the information demands posed by players on post-play visualizations and
the goals they pursue with such visualizations. We focused on three genres that enjoy great popularity within the competitive gaming
scene. Our results provide useful guideposts on which data to focus on by offering an overview of the relevance of different in-game
metrics across genres. Lastly, we outline high-level implications for the design of post-play visualizations.

Full text

What Players Want: Information Needs of Players on Post-Game Visualizations
GÜNTER WALLNER,
Eindhoven University of Technology, The Netherlands and Johannes Kepler University Linz,
Austria
MARNIX VAN WIJLAND, Eindhoven University of Technology, The Netherlands
REGINA BERNHAUPT, Eindhoven University of Technology, The Netherlands
SIMONE KRIGLSTEIN,
Masaryk University, Czech Republic, AIT Austrian Institute of Technology GmbH, Austria,
and University of Vienna, Austria
With the rise of competitive online gaming and esports, players’ ability to review, reect upon, and improve their in-game performance
has become important. Post-play visualizations are key for such improvements. Despite the increased interest in visualizations of
gameplay, research specically informing the design of player-centric visualizations is currently limited. As with all visualizations,
their design should, however, be guided by a thorough understanding of the goals to be achieved and which information is important
and why.
This paper reports on a mixed-methods study exploring the information demands posed by players on post-play visualizations and
the goals they pursue with such visualizations. We focused on three genres that enjoy great popularity within the competitive gaming
scene. Our results provide useful guideposts on which data to focus on by oering an overview of the relevance of dierent in-game
metrics across genres. Lastly, we outline high-level implications for the design of post-play visualizations.
CCS Concepts:
•Applied computing →Computer games
;
•Human-centered computing →Information visualization
;
Empirical studies in visualization.
Additional Key Words and Phrases: games, players, gameplay visualization, information needs
ACM Reference Format:
Günter Wallner, Marnix van Wijland, Regina Bernhaupt, and Simone Kriglstein. 2021. What Players Want: Information Needs of
Players on Post-Game Visualizations. In CHI Conference on Human Factors in Computing Systems (CHI ’21), May 8–13, 2021, Yokohama,
Japan. ACM, New York, NY, USA, 19 pages. https://doi.org/10.1145/3411764.3445174
1 INTRODUCTION
Video games strongly rely on players being able to make informed gameplay decisions which, in turn, requires
appropriate means to convey the necessary information to players. It is thus not surprising that information visualization
and video games share a long common past with visualizations taking on many dierent forms, ranging from simple
health bars to more complex representations such as mini maps or technology trees (see, e.g., [
8
,
54
] for an overview).
Historically, these visualizations were conned to a game’s user interface. However, the emergence of online-enabled
gaming devices made it possibly to unobtrusively and continuously collect large-scale behavioural data of players,
leading to new application areas for data visualization (cf. [
46
]). Especially within the context of games analytics,
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not
made or distributed for prot or commercial advantage and that copies bear this notice and the full citation on the rst page. Copyrights for components
of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on
servers or to redistribute to lists, requires prior specic permission and/or a fee. Request permissions from permissions@acm.org.
©2021 Copyright held by the owner/author(s). Publication rights licensed to ACM.
Manuscript submitted to ACM
1

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
visualizations have become an increasingly prominent tool to assist developers in deriving actionable insights from
the data [
47
]. This has lead to a wide variety of visualizations (e.g., [
2
,
3
,
14
,
16
,
23
,
48
]) being adapted or specically
developed for gameplay analysis.
While it can be observed that much of the current eorts are directed towards developers – see, for instance, the survey
of Wallner and Kriglstein [
49
] – players have been identied as another important target group [
8
,
49
]. Players constitute
a growing audience, also because of the rise of competitive online gaming and esports which spurred the interest
of players in post-play visualizations to reect and improve upon their in-game activities. Indeed, so-called training
visualizations have been identied as a primary purpose of visualizations in games besides status visualizations [
8
,
22
].
Training visualizations can be part of the game itself. This can take on the form of graphs showing progression over
time such as in StarCraft II (cf. [
22
]) or visual overlays such as optimal trajectories in racing games (cf. [
13
]). They can,
however, also be provided external to the games, for instance, by websites (e.g., [
25
,
43
]) or stand-alone tools (e.g., [
4
,
6
]).
In academia, however, player-centric visualization is only recently gaining traction with only a few approaches being
published to date (e.g., [
1
,
45
]). We argue that developments in this area are partly thwarted by the fact that there
currently exists limited knowledge about which information players rely on to make in-game decisions. Understanding
which information is important and why and which goals players pursue with visualizations is, however, critical to
create visualizations that satisfy players’ needs. Moreover, building an understanding of the information needs of players
cannot only contribute to training visualizations but can also be benecial for developing in-game user interfaces and
analysis approaches geared towards developers. Thinking further ahead, our results can also be useful for drawing
systematic comparisons between the needs of players and other potential users of visualizations such as spectators.
In an eort to narrow this gap in knowledge we thus conducted a study inquiring into the information needs of
players with respect to post-play data visualizations. As information needs can be assumed to vary across games, we
focused on dierent popular genres (real-time strategy, multiplayer online battle arena, and battle royale) in competitive
gaming to develop a broader view on the subject. The study consisted of interviewing players (
𝑁=
16) of games of the
above genres to explore and develop themes regarding the aims of using visualizations as well as important information
that should be visualized. Subsequently, the identied themes were quantied through an online survey (
𝑁=
264) to
establish the relevance of the themes across the three genres under investigation.
In summary, this paper contributes an overview of the goals players pursue with post-play visualizations along with
an inventory of important in-game information and their relevance across genres. Based on our results, we further
reect on a set of high-level recommendations for the design of training visualizations.
2 RELATED WORK
Visualizations of in-game data have found wide application within games user research and especially game analytics to
handle and make sense of large-scale datasets. The application of visualizations to support analytics has been discussed
by several authors (e.g., [
15
,
31
]). Wallner and Kriglstein [
49
] have charted the landscape of gameplay visualizations,
revealing a wide spectrum of dierent types in use, including spatial (e.g., [
15
,
48
]), temporal (e.g., [
2
,
38
]), and network
visualizations (e.g., [
28
,
33
]). While most research eorts in post-play visualizations within game analytics are directed
towards analysts, players have been identied as a main audience for visualizations as well [8,49].
Indeed, visualizations have fullled an important function in games since the early days of digital gaming by
conveying data about the in-game status to players on which they can base their decisions upon (cf. [
8
]). Bowman
et al. [
8
], using examples from a range of games, have developed a framework for the use of visualizations in games.
Similar reviews have also been conducted by Zammitto [
54
] and Medler and Magerko [
32
]. They identify a wide
2

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
array of purposes visualizations can fulll, ranging from in-game status visualizations to visual representations for
communication, progression, and training. Hazzard [
22
] makes a more high-level classication, distinguishing between
status visualizations and training visualizations serving the goal of improving player skill. In this paper, we are concerned
with the latter, particularly those who make use of tracked in-game data.
Indeed, training visualizations are garnering increasing popularity, partly fueled by the growth of the competitive
gaming scene and more and more developers making in-game data accessible through public APIs. This has resulted in
community eorts which utilize the available data to build analytics tools and websites. For instance, OpenDota [
43
],
DOTABUFF [
18
], and datdota [
12
] are data platforms for Dota 2 [
44
] that show various statistics. WOTInspector [
25
]
makes use of heatmaps to depict metrics such as damage or team movement as well as animated maps to visualize
data from World of Tanks [
52
]. Similarly, stand-alone tools such as Scelight [
6
] for StarCraft II [
7
] or the WoT-Replay-
Analyzer [
4
] for World of Tanks also heavily rely on visualizations to convey the data. Other tools combine replay
functionalities – that enable reviewing matches either directly in the 3D environment (e.g., [
42
]) or on 2D top-down
maps (e.g., [
5
,
39
]) – with dierent types of visual overlays depicting in-game telemetry data such as trajectories, cone of
visions, or kill locations. Usually such tools are game-specic but some such as Shadow [
39
] also oer a single solution
for multiple titles.
In recent years, research has also begun to develop and study player-centric visualizations that utilize telemetry
data. Wallner and Kriglstein [
50
] conducted a comparative evaluation of three visualizations – diering in terms of
abstraction level (detailed unit movements vs aggregated troop movement) – for the retrospective analysis of battles in
team-based combat games. Results indicated that there was no single visualization that was equally useful for all types
of analysis tasks performed by the players. Kuan et al. [
27
] designed a battle visualization for StarCraft II based on
a set of requirements players deemed useful in order to be able to learn strategies. While on overall the system was
appreciated by study participants they also pointed out several shortcomings, providing further evidence that satisfying
all player demands with a single visualization is challenging.
Research in information visualization has long stressed that understanding user tasks and information needs is
crucial for building suitable graphical representations that fulll them (cf. [
37
,
51
]). However, studies explicitly enquiring
into the information needs of players and their goals are still scarce. Karlsson [24] explored how players comprehend
information presented to them through in-game visualizations with results suggesting that non-gamers were more
attracted to dynamic status information while experienced gamers focused more on high-level overview information
such as the mini map. Afonso et al. [
1
], on the other hand, were interested how information is getting analyzed. Using a
spatio-temporal visualization of League of Legends [
36
] match data, they identied three distinct groups who analyzed
the data in dierent ways (e.g., players just describing the information presented, players making speculations based on
what is presented).
Lastly, albeit not having training as their primary purpose, visualizations to enhance spectatorship should be
mentioned here too as, for instance, watching streams of others people play has been associated with learning as well –
although the extent to which this is the case varied across studies (cf. [
21
,
41
]). Anyway, visualizations may provide
additional benets in this regard. For instance, Charleer et al. [
11
] evaluated two visual dashboards showing real-time
match data in addition to the games’ user interfaces. Their results indicate that they positively contributed to spectator
insight and experience but that careful design is required to alleviate information overload. Kokkinakis et al. [
26
]
developed an app that allows audiences to watch data-driven content alongside esports broadcasts. Evaluations of it at
two international Dota 2 tournaments revealed that spectators show propensity to engage with the data-based content.
3

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
As Charleer et al. [
11
], nding the right balance in terms of information density was identied as a major challenge.
Consequently, it can be expected that our ndings can also provide useful pointers for the design of such visualizations.
3 STUDY DESCRIPTION
The goal of this research was to explore information needs and to quantify their relevance across genres. We thus opted
for a mixed-methods approach to obtain a better understanding about which information players base their decisions
on and how this contributes to a) information needs with respect to post-play data visualizations and b) learning and
skill development in games. Specically, we rst conducted semi-structured interviews to develop recurring themes
which were then quantied through a follow-up online survey.
The whole study was approved by the ethical review board of the Eindhoven University of Technology. Interview
guidelines and survey were trialed with two participants each to ensure that the structure and questions asked were
appropriate for the intended audience and to pre-test timing and duration of both.
This paper focuses on the rst part (a) of this larger study, guided by three research questions. Firstly, understanding
information needs requires to understand why players view post-game data visualizations and the goals they pursue
with it. As such, our rst research question was:
RQ1: Why do players view post-game visualizations or replays?
Creating visualizations that are eective also necessitates to identify which data is important and meaningful for the
tasks to be achieved (cf. [
37
]), addressed by the remaining two research questions. As we are specically concerned
with visualizations for training we were interested in understanding which information players use to inform their
gameplay decisions and thus deem important for succeeding in a game, leading to our second research question:
RQ2: Which information do players rely on to make gameplay decisions?
Related to this, we were interested in what information players themselves would include in a graphical representation
to help them understand their in-game behaviour and thus perceive as relevant, thereby yielding the following research
question:
RQ3: Which information is considered relevant to be included in gameplay visualizations?
Please note that RQ2 focuses on the information players pay attention to during play while RQ3 focuses on data deemed
valuable posteriori as we assume that these may be dierent. For this study we focused on competitive online games,
covering three popular genres: real-time strategy (RTS), multiplayer online battle arena (MOBA), and battle royale
(BR) games. For the interviews (Section 3.1) we, however, restricted ourselves to one representative game from each
genre, selecting games that enjoy high popularity. These games (and genres) were chosen because visual summaries of
in-game data are regularly associated with games that rely on skill such as RTS and rst-person shooter games [
32
].
Popular data platforms for these games (e.g., [
6
,
34
,
43
]) also suggest a community embracing data-driven solutions for
analyzing in-game data. In the following we shortly describe the three games and the general characteristics of the
genre:
StarCraft II
[
7
] is a real-time strategy game (RTS) that revolves around three species competing against each other.
Starcraft II supports up to four players, with each player controlling a large number of units being led into battle
against the opponent(s). It has many of the genre’s typical features which include army management, base building,
and resource gathering. Players have to manage these aspects in order to emerge victorious. Base building is especially
important in StarCraft II to build a strong economy that supports the production of military units. Dierent RTS games
4

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
weigh the above features dierently and sometimes also include role-playing elements such as character leveling. They
are usually played from a top-down, usually isometric, perspective.
Dota 2
[
44
] is a multiplayer online battle arena (MOBA) game which means that each player controls a single character
within a team competing against a second team of players. Each team aims to win the game by destroying the main
structure of the enemy. Defensive structures such as towers can be constructed and computer-controlled units are
spawned periodically and roam the map. These can be killed for gold which, in turn, can be used to buy items or revive
the player character. MOBA games are characterised by a large number of playable characters, each having unique
strengths, weaknesses, and abilities. They are usually played on a map from a top-down perspective like RTS games
and put strong emphasis on team play and a well-balanced and eective team composition.
PubG
[
35
] is a battle royale game (BR) which involve up to hundreds of players. Each player controls a single character
with the goal to eliminate all other players and be the last one alive. Players start with no or limited gear and need to
search the environment for equipment and weapons. Every few minutes the map shrinks, resulting in a more conned
playable area. Players do not respawn once they have been killed. BR games are usually played in rst or third-person
perspective and follow in general the above gameplay but often also implement dierent variations.
3.1 Part 1 – Interviews
We rst explored our research questions using semi-structured interviews. To recruit a diverse group of interviewees
who cover a wide range of experiences and views and to ensure that basic criteria were met, we prepared a screener
to identify potential interview candidates. The screener asked about demographics, self-reported gaming experience,
familiarity with visualizations, as well as for which game (Starcraft II, Dota 2, PubG) they would like to participate.
The screener was distributed among relevant subreddits and Discord servers for the selected games as well as the
private network of the second author. From 30 responses to the screener we, in the end, selected 16 candidates (15
male, 1 preferred not to reveal their gender) for being interviewed: seven Dota 2, six Starcraft II, and three PubG players.
Age ranged from 20 to 29 (mean = 23.75, SD = 2.8) with Starcraft II players being slightly older (mean = 25.3, SD =
3.3) than Dota 2 (mean = 23.6, SD = 1.9 ) and PubG (mean = 21, SD = 1.0) players. All of them had multiple years of
gaming experience and were familiar with at least dierent types of charts and diagrams as well as heatmaps. Two
participants self-identied as ex-professional Dota 2 players. While one now plays for relaxation only, the other one
acts as a role model within the Dota community and serves in captain-like roles. Further two participants were former
semi-professional players (one Dota 2 and one Starcraft). Both described themselves now as more casual players but
still maintain connections to the community. Of the 16 participants, four highlighted that they are active members
of the game community by acting, for example, as commentator, coach, role model, or by running a Discord server.
Six participants indicated to the play the games in a more socially oriented way at the time of the interviews while
three participants considered themselves highly competitive players. Each of the three games was represented by both
social and competitive players. In the following, we will refer to them as
𝐷
1
−𝐷
7,
𝑆
1
−𝑆
6, and
𝑃
1
−𝑃
3, with the letter
indicating the game.
The interviews lasted for about an hour and were conducted via Discord or Microsoft Teams. At the start of the
interview participants were introduced to the study’s goals and consent to audio record them was obtained. The
interview itself was semi-structured and covered the game itself (e.g., why it is played, what they like or dislike about
it, which activities they engage with), skill development and learning, decision making, the information they use for
that, and the role visualizations play. Recorded interviews were transcribed and analyzed using inductive iterative
thematic analysis [
9
] with respect to our three research questions, that is 1) why players use visualizations of game
5

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
data or watch replays in order to understand for which purposes they employ them, 2) the information they use to
make gameplay decisions, and 3) how relevant they deem dierent types of information to be visualized. In a rst
step, the interview transcripts were coded to derive an initial set of themes across all interviews. These were rened
based on discussion with a second researcher, followed by re-coding and a second round of discussion to ultimately
construct the nal themes and higher-level themes. Finally, the interviews were coded according to these. While we
interviewed players from dierent games to approach the topic from dierent perspectives we paid attention to derive
genre-independent categories to increase generalizability. The relative importance for the dierent genres was assessed
using the online survey. The nal set of themes together with a short description, the number of mentions, and the
number of participants mentioning them, are summarized in Table 1to Table 3.
3.2 Part 2 – Online Survey
To quantify the themes developed in Part 1 and assess their importance for dierent genres we administered an online
survey through Google Forms. The survey was advertised on subreddits and Discord servers for popular competitive
games belonging to the three selected genres, especially for Starcraft II [
7
], Age of Empires [
19
], Command & Conquer [
53
],
League of Legends [
36
], Dota 2 [
44
], Fortnite [
20
], and PubG [
35
] as well as some general gaming related subreddits. The
survey included basic demographics such as age and gender, previous gaming experience in years, and hours spent
playing in the last 30 days prior to taking the survey. Participants also had to indicate the game for which they were
taking the survey, with the above mentioned games being available as preset answer options. However, participants
could also specify another game. The main part of the survey involved rating the themes inferred through the thematic
analysis on 7-point rating scales with labeled endpoints. Generally, one theme was represented by one item in the
survey but a small number of themes was split into sub-questions to assess dierent aspects of them. For instance, Kills
& Deaths was split to inquiry about their temporal and spatial dimension separately.
In total, we received 292 survey responses of which 15 were excluded because they were lled out with respect to
a game not falling withing the three explored genres. This was followed by a longstring analysis to detect careless
responders with repeated answering schemes (cf. [
30
]). This resulted in 13 additional cases to be excluded, yielding a
total of 264 responses for analysis. Of those, 249 were male, 10 female, 1 non-binary, and 4 preferred not to disclose
their gender. In terms of age, most participants (
=
153) were in the range of 20 to 29 years, 77 were below 20 years, and
34 were above 29 years. Distribution across the three genres was well balanced with the BR and MOBA genre receiving
almost the same number of responses. Specically, 83 participants responded with respect to BR games (Fortnite = 60,
PubG = 23) and 80 with regard to MOBAs (League of Legends = 58, Dota 2 = 21, Heroes of the Storm = 1). The RTS genre
received slightly more responses, 101 in total (Starcraft series = 67, Age of Empires series = 31, Command & Conquer
series = 3). In terms of gaming experience, participants played games on average since 5.5 years (SD = 4.3, min = < 1
year, max = 20 years) and played on average 63.6 hours (SD = 54.2, min = 0, max = 350) in the 30 days prior to taking
the study.
Likert-scale responses to the survey were treated as ordinal and thus analyzed using non-parametric tests. In order
to assess dierences between the three investigated genres we calculated Kruskal-Wallis tests for each category (i.e.
question). Due to the number of comparisons we calculated adjusted p-values for these tests to control for false
discovery rate (Benjamin-Hochberg procedure). If the test was signicant, it was followed by Dunn’s post-hoc test with
a Bonferroni corrected alpha level of .016. Figure 1shows the results with respect to why players rely on post-game
data visualizations or watch replays. Figure 2depicts which type of information about themselves or the opponents
6

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
Table 1. Themes regarding the reasons for watching replays and viewing post-game data visualizations and summaries, along with
the number of participants mentioning it (#P) and the number of statements (#S).
Theme/Category Description #P #S
Impact of Decisions
insights on the eects of decisions and their inuence on the game, if they
were the right decisions or whether others would have been better
8 12
Efficiency improving eciency with respect to
Resources & Items gaining insights on the ecient use of resources and items, including
the timing of use as well as the optimal choice of items and on what
to spend resources on
9 12
Seqencing
gaining insights with regard to sequencing of items, buildings, levels &
upgrades, etc.
3 5
Movement movements and how important it is 3 3
Understand Situations
viewing of specic situations that the player did not understand or was
surprised by
7 7
Predictability being able to predict and disguise behaviour
Being Unforeseeable to understand how to be less predictable for others 5 7
Anticipate Enemy to be able to better anticipate and predict enemy plans and actions 4 6
Learn from Others
watching how other people play to understand how well they do and what
actions they took as well as identifying new ways of playing
6 7
Learning
identifying learning points and mistakes, gaining insights in how much
dierent factors inuence the game, and picking what to focus on to
improve next
5 6
Personal Satisfaction viewing points, score, gold, damage dealt, etc. for personal satisfaction 2 2
players nd useful for making gameplay decisions. Finally, Figure 3examines how relevant players deem the inclusion
of certain information in a visualization.
4 RESULTS
In the following we discuss our results based on the three research questions outlined in Section 3. For the sake of
presentation, we will designate median scores
<
4as low,
≥
6as high, and as moderate otherwise. Detailed statistics
can be found in Figure 1to Figure 3. Categories derived from the thematic analysis of the interview transcripts are
written in small caps.
4.1 Why do players view post-game visualizations or replays (RQ1)
All except one category (cf. Table 1) fall under the larger umbrella of players wanting to understand and improve upon
gameplay, showing the high relevance of post-play visualizations for training and learning purposes. Interviewees
commented about replays and visualizations being valuable for Learning about the game in general, especially also by
being able to Learn from Others by observing how they play, what mistakes they make, and potentially identifying
new ways of playing. Interviewees expressed interest to review the Impact of Decisions to get an impression how their
choices inuenced the game. Participants also appreciated replays and visualizations for Understanding Situations
which were not obvious during play. This is nicely captured by the following comment from
𝑆
3:Sometimes you lose one,
and if I don’t get why I lost, I check the replay and try to gure that out. These aspects were all rated highly relevant in
the survey across all three genres (see Figure 1), conrming the impressions gained from the qualitative analysis.
Players were also especially concerned with improving their Efficiency within a game by deducing ways to better
use Resources & Items; to learn about Seqencing, that is, in which order and when to use upgrades or construct
7

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
†Kruskal-Wallis H test signicant at false discovery rate adjusted 𝑝<.05, Dunn’s post-hoc test for multiple comparisons signicant at
∗𝛼=.016 (Bonferroni corrected), ∗∗𝛼=.01,∗∗∗𝛼=.001
Fig. 1. Reasons for watching replays or viewing post-game data visualizations (1 = not relevant, 7 = very relevant) across the dierent
genres.
buildings; and to improve Movement, including the positioning of units. The following quote exemplies the general
importance of eciency well: I would like to see when I could have done something faster or more ecient [
𝑆
4]. While
Movement was considered relevant for all three genres, Seqencing was rated signicantly more relevant in case
of RTSs and BRs compared to MOBAs, while Resources & Items showed signicant dierences between RTSs and
MOBAs.
Another frequent reason for referring to visualizations and replays was players’ desire to better Anticipate Enemy
behavior and thus being able to predict future actions while at the same time also gaining insights on how to disguise
their own undertakings to be more unforeseeable. For example, the following comment by P1 reects on the former
aspect:
Especially the death cam is important. It tells you how and why you were vulnerable but also gives insight in
how the enemy killed you, so you can try to learn from their thinking and technique.
Learning to Be Unforeseeable was especially considered useful in the case of BR games but was only considered
moderately relevant in case of MOBAs and RTSs. Being able to anticipate the enemy was rated more relevant in case of
RTSs and BRs than for MOBAs.
Lastly, interviewees made remarks about how watching replays can serve as Personal Satisfaction by seeing how
well oneself succeeded or others were failing, but those were few. This was also evident in the survey responses were
participants rated it of lower relevance compared to all the other aspects related to learning about dierent aspects of a
game.
4.2 Which information do players rely on to inform gameplay decisions (RQ2)
Table 2lists the dierent types of information players considered necessary for making informed gameplay decisions.
Participants concluded that both information about oneself (or the team one is being part of) and about the opponent(s)
are important for making gameplay decisions.
In general, players considered the same type of data important to know about both sides. Knowing the Positions &
Movement about oneself (and one’s team) as well as about the enemy was the most mentioned type of information to
8

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
Table 2. Categories of which information is important when making decisions during play, along with the number of participants
mentioning it (#P) and the number of statements (#S).
Theme/Category Description #P #S
Opponent information important to know about the opponent
Positions & Movement
the importance of knowing enemy movement and positioning 14 19
Buildings
information about enemy buildings and its importance for making deci-
sions
4 13
Play Style
how the enemy playstyle (e.g., aggressive, defensive) inuences one own’s
playstyle
8 12
Entity
∗
Characteristics unique and specic characteristics about the race, hero, or build the enemy
is playing and how it inuences decision making
6 7
Equipment
enemy equipment, weapons, items, abilities, or similar that provide options
and inuence play
4 6
Levels
the importance of levels of enemy units, heroes, buildings etc. for decision
making
3 3
Resources how the resources the enemy has available inuences decisions 3 3
Hit Points how enemy hit points or health inuences decision making 2 2
Kills & Deaths how enemy kills and deaths inuence decisions 2 2
Start Location how the enemy starting location can inuence decision making 1 1
Team Role how the roles in the enemy team inuences decision making 1 1
Myself & One’s Team information important to know about the player (or the player’s team)
Positions & Movement
how the player’s and/or team’s positioning and movement inuences
decisions and how to proceed
7 8
Chat
how information provided through the team chat is important for decision
making
5 5
Entity
∗
Characteristics
the importance and inuence of the chosen hero, race, build, and units
(and their specic characteristics) on play and how decisions are made
4 4
Equipment
how the player’s or team members’ equipment, abilities, weapons, and
items inuence play
3 3
Resources
the importance of the player’s and/or the team’s resources on decision
making
2 2
Levels
how levels of, e.g., buildings and abilities decide which decisions are made
3 3
Team Role
how the role of the player in the team inuences the plan and actions to
be taken
3 3
∗Note: Entity is being used as a generic term for dierent playable characters such as heroes, units, and more high-level constructs which
can have unique characteristics as well. Categories written in italics have been mentioned for oneself and the opponent.
inuence decisions. D3, for instance, noted that generally, enemy movement is very important combined with your own
position in the game. D7 elaborated further on this, saying:
See where people are walking, and deduce what their plans are. Figure out their goal, what and why they
want to achieve. Then anticipate if/how you can stop this [...]
This also showed in the ratings, with participants rating it of highest importance across all genres (cf. Figure 2).
Players also made frequent references about how entity characteristics
1
and their specic Levelling inuence
the choices they make. While both categories are about playable entities we distinguished between the two, to highlight
that players consider both the static characteristics that entities possess while also being inuenced by their evolving
attributes through continual leveling. For example, D1 commented:
1Entity is used here as a generic term for all dierent types of playable characters (e.g., units, heroes, race) within a game.
9

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
†Kruskal-Wallis H test signicant at false discovery rate adjusted 𝑝<.05, Dunn’s post-hoc test for multiple comparisons signicant at
∗𝛼=.016 (Bonferroni corrected), ∗∗𝛼=.01,∗∗∗𝛼=.001
Fig. 2. Importance (1 = very unimportant, 7 = very important) of dierent types of information about the player itself or the own
team (le) and about the enemy (right) in order to make gameplay decisions across the dierent genres.
Heroes that I know are good against me inuence me mostly. When I play Juggernaut [a hero in Dota 2], for
example, I know there are only a few heroes that can really damage me. Based on that, I can decide where I
am safe or what I can reasonably win.
Both showed to be most relevant in MOBAs and RTSs, with players of BR games only attaching low (maximal moderate)
relevance to it. Related to this, participants reected on how their Eqipment (e.g., weapons, items) and that of the
opponent has implications on their in-game interactions, as illustrated well by the following comment from D1:
Other than that, obviously your items, gold, and weapons are important; those decide your options and what
lines of play you have available.
While generally considered of moderate to high importance, ratings showed signicant dierences between the genres,
with respondents considering it more important for MOBAs and BRs than for the RTS genre. The Resources one or the
contender(s) have at their disposal were also mentioned as having an eect on the decisions being made. Resources
were considered as having moderate to high importance in all the genres with dierences in the ratings (except one)
being statistically insignicant. Lastly, interviewees elaborated – although rarely – on how the role of the player or the
roles in the opponent’s team aects their choices. The quantitative results also only attested high relevance to it in case
of MOBAs, with scores for RTS and BR games (low to moderate) being signicantly lower.
How information provided through the Chat inuences players was only mentioned with reference to the own
team which is reasonable because the chat of the other team is not accessible during gameplay. For example, D7 noted:
10

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
What your teammates are saying is also very important; you can’t see everything yourself, mostly there’s
someone coordinating (like me) who provides information about what to watch out for, and where the enemy
is moving and what we should do.
The communication between team members was considered a highly important source of information in case of MOBA
and BR games, while it was considered of signicantly less (albeit still moderate) relevance for RTS games.
It is important to highlight that throughout the interviews participants focused more on which opponent-related
information they rely on to inform their decisions. Especially, knowledge about the enemy’s Buildings was a reoccurring
theme, particularly in case of RTS players as it is considered important for taking appropriate counter measures or for
deciding when to attack. Survey responses conrmed the high relevance in case of RTSs but also showed that it can
be moderately valuable for making decisions in MOBAs and BRs. Players also indicated to pay attention to the Hit
Points (i.e. the amount of damage a character can withstand) or health the enemy has, with players indicating that this,
for example, decides whether to engage or retreat. While not mentioned often in the interviews, survey respondents
conrmed its high importance for MOBAs and BRs but deemed it only of moderate relevance for RTS games. Start
Locations as well as locations of Kills & Deaths caused by and of the enemy were also noted, albeit very rarely. The
quantitative results conrmed this, with both being rated to be only of low to moderate importance, with the Kills &
Deaths in case of MOBAs rated highest. Lastly, interviewees made frequent references to the general Playstyle of the
opponent, for instance, if the enemy follows a defensive are aggressive strategy or shows signs of greediness. Playstyle
also formed, together with Positions & Movement and Resources, the most highly rated category independent of
genre.
4.3 Which information is considered relevant to be included in gameplay visualizations (RQ3)
Asked about which information players judge relevant to be included in a post-game visualization, interviewees made
recurring references to the visualization of specic data as-is (e.g., kills, scores, use of resources) but also discussed
the relevance of visualizing higher-level knowledge (e.g., map control, ghts, being in advantage) that requires the
integration and combined analysis of various low-level data (e.g., individual positions, death locations, character levels).
Table 3summarizes the major themes arising from the interviews with high-level concepts written in boldface. Figure 3
presents the scores for the dierent categories across the three genres.
Some of the emerged categories relate directly to the information players rely on during gameplay (in particular
Movement,Resources,Kills & Deaths, and Build Order). As during play, information about Movement and
Resources proved to be the most discussed aspects to be visualized, reinforcing their importance for succeeding in
the investigated games/genres. Information on movement about oneself or the opponent was rated of moderate to
high importance with no signicant dierences between the genres. Resources although also scoring in the range
from moderate to high were, however, considered of signicantly higher relevance in RTS games compared to BRs and
MOBAs.
Interviewees also suggested to include details about Item Use and about general Scores such as points, position in
the leaderboard, and similar to see how well one is doing in the game (also in relation to others). Both, however, were
considered of moderate to low importance with scores also varying signicantly across the genres. Kills & Deaths
(both in terms of location and time) and the Number of Units were also found worth visualizing with the former
being moderately important for BR and MOBA games but signicantly less so for RTSs whereas the latter showed the
opposite pattern, being considered highly important for RTSs but unimportant for MOBAs and BRs. Information about
11

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
Table 3. Themes with respect to the visualization of data for post-play analysis. Categories in boldface denote higher-level concepts
requiring the integration of several data types. #P indicates the number of participants and #S the number of statements.
Theme/Category Description #P #S
Movement
visualizing the movement of heroes and units across the map, also in
relation to other relevant statistics
9 20
Resources
visualizing acquisition and use of resources, also in relation to other sta-
tistics and events, to gain insights regarding their optimal use
9 20
Fights
times when important ghts happened, specically combined with other
relevant statistics to be able to see how these inuence a game
4 8
Item Use
suggestions to visualize the timing of item use, also in relation to other
statistics and events
5 8
Scores
visualizing scores, points, rankings, or similar to be able to see who is
ahead at which moment, why, and how this relates to other statistics
7 8
Map Control
suggestions on visualizing map control, i.e. which team controls which
parts of the map over time, and where the front line is
4 7
Kills & Deaths visualizing kills and deaths and their impact on the game 4 7
Blocking of Supplies
observing time spans when resources are being cut o, to see how this
impacts the game and how long it takes to recover from it
3 5
Number of Units army size and/or the number of workers 3 5
Levels & Abilities
seeing when leveling/upgrading is initiated and nished and when abilities
are available, when they are used and when they are available again
3 5
start Location
viewing the enemies’ starting location(s) in relation to other statistics (e.g.,
win rate) to understand how this aects the game
3 5
Pressure
suggestions to quantify, process, and visualize the amount of pressure
applied to and by an enemy to gain insights in how this inuences games
and winrates
3 5
Comparison
visually comparing one’s own game, to create reference and give statistics
a semantic meaning
Player Base to compare oneself with how other players perform on average in
relation to specic aspects
4 5
Personal Progress to track the improvement with respect to one’s own progress and
learning curve
2 3
Optimum or Goal to compare to a set goal or the optimal way of playing, sequence, etc. 1 1
Match Fairness
suggestions for quantifying and visualizing the fairness of a match, to help
understand reasons for winning and losing and help people to be at peace
with losing
2 4
Build Order order and timing of constructing buildings 3 4
Damage the amount of damage dealt to or by specic characters, units, or teams 3 4
Field of View
suggestions to visualize how the own’s and enemy’s eld of view moved to
be able to gain insights regarding eciency as well as which information
was missed because one did not observe a specic area of interest at the
right time
3 4
Being in Advantage
processing and visualizing moments where one team or player has a
distinct advantage over the other(s), if this is noticed and if it is utilized
properly
1 3
Levels & Abilities was considered of moderate importance for RTS and MOBA games with both showing signicant
dierences to BRs for which it was considered irrelevant. In contrast, the Start Location was of high importance
for BRs with signicant dierences to RTSs and MOBAs (very low relevance). Visualization of Build Orders was
considered highly relevant for RTS but signicantly less so for the other two genres. Damage dealt or received emerged
as another important source of information, being most important for BRs, followed by MOBAs, and lastly RTSs with
12

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
†Kruskal-Wallis H test signicant at false discovery rate adjusted 𝑝<.05, Dunn’s post-hoc test for multiple comparisons signicant at
∗𝛼=.016 (Bonferroni corrected), ∗∗𝛼=.01,∗∗∗𝛼=.001
Fig. 3. Relevance (1 = not relevant, 7 = very relevant) of visualizing certain types of information for gaining new insights, learning,
and/or being more motivated and confident across the dierent genres.
signicant dierences between the scores. Players also reected upon the value of visualizing the Field of View in
order to be able to assess which areas were observed or not, potentially causing them to miss important information. It
was rated of moderate to high importance (for BRs) but the dierences were not signicant.
In addition to players reecting upon the visualization of data items, the interviews also revealed an interest in
visualizations that not only show the data ’as is’ but which rather show higher-level knowledge extracted from the data.
For instance, S6 reected on how visualizations can help to draw attention to mistakes made:
Yeah with pure data, you basically ask the gamer to make conclusions by themselves, I think that is the
bottleneck. I think analyzing data and drawing mistakes from patterns, can really help. It’s about highlighting
the big mistakes. That is hard though, since there are so many elements contributing to the big picture.
Among those, the most recurring concept in the interviews was about conveying details about Fights and how they
inuenced a game. Fights were also considered highly relevant by the survey respondents across all genres, of which
MOBA players rated it the highest. Related to this, the interviews revealed a tendency to be able to observe who controls
which parts of the map during which time periods and how the front line changes, i.e. Map Control. The ratings
conrmed the high importance of this kind of information for MOBA and RTS players. D1 mused about how current
visualizations lack in this aspect:
However, even with all the visualizations used in coverage of professional matches, I have never seen something
like that, using front line, zone of control, map control, etc. you have to translate that from the map yourself,
a little. Visualizing this could really provide great insights – I think.
13

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
Related to this, players also considered it relevant to understand the Blocking of Supplies and the consequences
arising from it, with RTS players considering it highly relevant. Dierences to MOBA and BR players who rated it
of moderate and low importance, respectively, are signicant. The concept of Pressure also arose repeatedly in the
interviews. Players commented on how important it is to apply pressure to the enemy and how they would appreciate a
way to quantify and visually convey the amount of pressure. Survey ratings also attested moderate to high importance
to it with no signicant dierences between the genres.
Two participants also made suggestions for visualizing the Match Fairness to help understand why a match might
have been lost or won. Survey respondents thought it to be of moderate relevance in all three genres, with no signicant
dierences between them. Showing if someone has a distinct advantage (Being in Advantage) during a match –
although only suggested by one participant to see, for instance, if the advantage was used or missed – was considered
of high importance for all three genres by the survey respondents. Lastly participants made recurring statements
specically about visualizations that would allow them to make Comparisons to situate their performance either (in
order of decreasing number of mentions) with respect to the general Player Base, their Personal Progress, or to an
Optimum or Goal. Despite the varying relevance in the interviews, survey respondents agreed that all three types of
comparisons are valuable (moderate to high) for post-play analysis with no pronounced dierences between the genres.
5 DISCUSSION
Our results provide an overview of which goals players pursue with post-play visualizations (cf. Table 1), which
information is relevant for them for decision making (cf. Table 2), which information to visualize (cf. Table 3), and how
their importance varies across three genres (see Figure 1to Figure 3).
The ndings show that players of competitive games do have an interest in using post-play visualizations of in-game
data as well as replays to help them learn about a game and develop new skills. Viewing them for personal satisfaction,
on the other hand, was of comparatively lower relevance. Learning has also been shown to be a reason for watching
streams [
21
] but Sjöblom and Hamari [
41
] found that learning about game strategies is of subordinate importance,
hypothesizing that the live aspect could be detrimental in this regard. Visualizations are dierent in this sense as they
can be viewed and manipulated by the users themselves, enabling learning about one’s own specic interests and at
one’s own pace. While players valued visualizations to learn from their behaviour and from others in general, our
analysis also revealed more specic themes revolving around situational understanding, awareness of the consequences
of decisions, a strong focus on improving eciency in various aspects, as well as predictability. While most reasons
were common across all three genres, some also varied across them (cf. Figure 1). This shows the importance of tailoring
training visualizations to the specic and also varying information demands of players. For example, a visualization to
convey the impact of decisions will need dierent design considerations than a visualization aiming to help improve
eciency.
As with the reasons for using visualizations, certain information players rely on during play and would like to see
visualized was considered equally important while the relevance of other information varied across genres. This can
partly be explained by some aspects being highly game specic (e.g., build orders to quickly establish a strong economy
are quite specic to RTS games) and partly by the number of units involved as well as the importance of individual
unit characteristics and the level of cooperation needed between players. For example, players of RTS games need to
command a large number of units where individual unit characteristics take on less importance while MOBAs put
strong emphasis on character attributes and leveling of the heroes which also have to complement each other well.
This places dierent emphasis on the level-of-detail of the data to be displayed.
14

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
5.1 Implications for the Design of Post-Game Visualizations
In the following we reect on high-level implications for the design of post-game visualizations, in particular visualiza-
tions for training, skill development, and learning as these formed the main reasons for using such visualizations.
5.1.1 Careful choice of data. With respect to the data players rely on during play (see Table 2), it is evident that, as
mentioned above, some data has general importance across all three genres (cf. Figure 2). The most relevant being
information on movement which can be attributed to the highly spatial nature of the games under investigation.
However, other information varied signicantly across the genres which is to be expected as their gameplay mechanics
are dierent. This places emphasis on deliberately choosing the data to be included in a visualization in order to keep it
simple, avoid visual clutter, and to make it useful for players of the respective games. As with visualizations part of a
game’s user interface [
10
] – or visualizations in general for that matter – post-play visualizations need to carefully
consider the information they present to neither overload the player with nor miss out on important information. Our
results provide guideposts on which aspects to focus on for the dierent genres (cf. Figure 3).
5.1.2 Choice of visual representation. Once the data has been selected appropriate representations for the data have
to be chosen, which is inuenced by the data to be presented and the task to be supported. The goals summarized in
Table 1and relevant information for players listed in Table 2and Table 3can be helpful in this respect. For example, a
visualization focused on eciency may put stronger emphasis on the temporal aspects of the data while a representation
of map control may attach greater weight to spatial relationships. Besides, our results can be valuable for choosing the
appropriate level of detail for representing the data (i.e. individual or aggregated). Revisiting the RTS example above,
the lower importance of individual unit characteristics paired with the large number of units may make, for instance,
aggregation of movement data an acceptable option.
Lastly, while we did not systematically code for it, we would like to – also because we feel it received comparable less
attention – raise another issue which surfaced anecdotally in the interviews, namely that the represented information
needs to be reliable. This is, for example, reected in the quote of S6 in Section 4.3 who raises the point that making
inferences oneself can be prone to errors or in the following statement from P1: I also question how realistic it is
in how data is interpreted. Indeed, Wallner et al. [
48
], proposing a developer-centric visualization, found that game
developers started to question the accuracy of the evaluated visualization when the depicted data did not match the
player behaviour they expected. In short, players need to have condence that the displayed data is represented in a
way that is trustworthy (see [
29
] for an overview of the dierent dimensions of trust in information visualization) and
that they can rely on it to make correct inferences with respect to the goals of the analysis.
5.1.3 Take advantage of data not accessible during gameplay. In general, players discussed information about themselves
(or the team they are part of) and information about their opponent(s) with the latter receiving slightly more attention.
When reecting on decision making players may implicitly assume certain information as given as they are constantly
aware of it during gameplay. As such they thus might not have explicitly stated it which could partially explain why
information about oneself resulted in fewer categories than about the enemy. While there is a large overlap between
the themes, some where unique to either of them. Some of the dierences may also stem from the fact that not all
information is accessible to players during play such as the chat of the enemy team. Post-game visualizations are,
however, not restricted to the information visible during play and should take advantage of being able to include data
otherwise inaccessible. This can increase situational understanding and benet the assessment of the impact of decisions
15

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
(which have been shown to be important reasons for reverting to visualizations and replays) because cause-eect
relations between the own and enemy behaviour can be made apparent.
5.1.4 Information becomes more/less relevant in retrospect. Our analysis also showed an overlap of themes between
the information players rely on during gameplay and they deem relevant to be included in visualizations, which is
to be expected. However, it also hinted at the fact that the relevance of certain information changes. For instance,
enemy movement was considered more relevant while playing whereas the movement of the player itself gained more
relevance in post-play visualizations (compare Figure 2with Figure 3). That is, post-game visualizations should consider
the fact that certain information may become more important retrospectively.
5.1.5 Convey high-level concepts. In terms of the type of information players consider relevant, the emerged themes
not only showed on interest in specic ’low-level’ data (e.g., number of units, kills and death locations) but also in
’high-level’ concepts (e.g., pressure, map control, ghts) which require the integration of dierent data types. For
example, ghts are not only characterized by the individual death locations but also who shot at whom, from where
units joined a ght and where they retreated to, and similar. Our results indicate that these high-level concepts were
generally rated highly relevant across all genres, while the value of dierent types of individual data varied strongly
between them. That is, training visualizations should move beyond visualizing individual metrics but rather combine
them in a way that conveys high-level knowledge that relates to macro gameplay concepts. This is not to say, however,
that such visualizations should completely neglect the representation of low-level data. Rather, it should be seen as
an opportunity to carefully combine both aspects to help players understand how a series of small decisions impacts
gameplay on a higher level. In that sense, such visualizations could benet by including articial intelligence and data
analytics methods to process and extract these high-level information from the individually tracked in-game metrics.
5.2 Limitations & Future Work
Our study was exploratory in nature involving a large number of statistical tests. While we took measures to control
for the number of comparisons the results should still be regarded as exploratory rather than conrmatory. In this
sense, signicant results should be interpreted as cases where visualization designers should be particularly mindful of
dierent game genres having dierent information needs and priorities.
In our analysis we focused on three popular genres to cover a variety of gameplay mechanics occurring in competitive
games. While it can be expected that our results, at least partly, transfer to others genres as well (e.g., rst-person
shooters), future work needs to conrm our results across further genres. Our focus on three genres also means that
the list of goals and relevant information for players may not be exhaustive. In that sense, further studies with other
games will help to build a more comprehensive inventory of player goals and information needs.
Dierent user groups with their dierent interests and requirements (e.g., developers using analytics to improve
game design and balancing, players to improve performance, spectators in being able to eectively follow gameplay)
will also likely have varying information needs. However, not much systematic knowledge about the needs of the
dierent groups currently exists. As such, our study can contribute and be useful for understanding dierences between
dierent types of users by conducting similar studies with other user groups in the future.
In this paper we focused on the information players would like to see rather than on the specics characteristics of the
data, for instance, if the data is spatial or temporal and which specic visualization techniques should be used to visualize
it. This was a deliberate choice as the specic attributes of the information can dier from game to game. However,
multiple taxonomies for choosing appropriate visual representations for specic data types have been proposed in
16

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
information visualization (e.g., [
40
]), games (e.g., [
49
]), and related elds such as competitive sports (e.g., [
17
]) to date.
These taxonomies can be useful in our context as well and hence we considered it out of scope of this work.
With many analytics tools being already available to players (see Section 2), an interesting avenue for future research
would be to evaluate existing analytics tools with respect to the information needs and design implications suggested in
this paper. While some of the identied dierences in genres can be explained by the dierent genre-specic gameplay
mechanics other dierences may require further inquiries to better understand why certain information is more or
less useful. Future work can also build upon our results to build new post-play visualizations, specically training
visualizations, that align with players’ goals and information needs. Evaluating these will bring along new challenges
since assessing training progress and skill development will require long-term studies that are carefully controlled for
confounding factors.
6 CONCLUSIONS
Visualizations for the respective analysis of gameplay have gained increased interest among players, especially those
that are interested in competitive gaming. However, little is known about the specic aims players pursue by reviewing
such visualizations and which information is deemed relevant for them to be included. Our study has shown that post-
game visualizations are mostly reviewed for training purposes, compared to replays where learning is of subordinate
importance [
41
]. By comparing information demands across three popular genres we established an inventory of
relevant information for them. Paired with the specic analysis goals and the high-level implications arising from our
analysis this can serve as a practical reference for the design of goal-directed and meaningful post-play visualizations.
Future work may, however, investigate other genres to expand upon this initial inventory. With player-centric
visualization still being a new research area, we hope our ndings spark inspiration for creating visualizations that
address tasks not yet or only marginally covered by existing gameplay visualizations.
REFERENCES
[1]
Ana Paula Afonso, Maria Beatriz Carmo, Tiago Gonçalves, and Pedro Vieira. 2019. VisuaLeague: Player performance analysis using spatial-temporal
data. Multimedia Tools and Applications 78 (2019), 33069–33090. https://doi.org/10.1007/s11042- 019-07952- z
[2]
Shivam Agarwal, Günter Wallner, and Fabian Beck. 2020. Bombalytics: Visualization of Competition and Collaboration Strategies of Players in a
Bomb Laying Game. Computer Graphics Forum 39, 3 (2020), 89–100. https://doi.org/10.1111/cgf.13965
[3]
Sabbir Ahmad, Andy Bryant, Erica Kleinman, Zhaoqing Teng, Truong-Huy D. Nguyen, and Magy Seif El-Nasr. 2019. Modeling Individual and
Team Behavior through Spatio-Temporal Analysis. In Proceedings of the Annual Symposium on Computer-Human Interaction in Play. Association for
Computing Machinery, New York, NY, USA, 601–612. https://doi.org/10.1145/3311350.3347188
[4] Aimdrol. 2020. WoT-Replay-Analyzer. https://github.com/Aimdrol/WoT-Replay- Analyzer Accessed: August, 2020.
[5] Bang & Jensen I VS. 2020. Noesis - Interactive CS:GO analytics. https://ww w.noesis.gg/ Accessed: November, 2020.
[6] Andras Belicza. 2014. Scelight. https://sites.google.com/site/scelight/ Accessed: August, 2020.
[7] Blizzard Entertainment. 2010. StarCraft II. Game [PC]. Nintendo, Irvine, CA, US.
[8]
Brian Bowman, Niklas Elmqvist, and T. J. Jankun-Kelly. 2012. Toward Visualization for Games: Theory, Design Space, and Patterns. IEEE Transactions
on Visualization and Computer Graphics 18, 11 (2012), 1956–1968. https://doi.org/10.1109/TVCG.2012.77
[9]
Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology. Qualitative Research in Psychology 3, 2 (2006), 77–101. https:
//doi.org/10.1191/1478088706qp063oa
[10]
Josh Bycer. 2014. Informed Decisions and their role on Game Design. https://www.gamasutra.com/blogs/JoshBycer/20141217/232510/Informed_
Decisions_and_their_role_on_Game_Design.php Accessed: August, 2020.
[11]
Sven Charleer, Kathrin Gerling, Francisco Gutiérrez, Hans Cauwenbergh, Bram Luycx, and Katrien Verbert. 2018. Real-Time Dashboards to Support
ESports Spectating. In Proceedings of the 2018 Annual Symposium on Computer-Human Interaction in Play. Association for Computing Machinery,
New York, NY, USA, 59—-71. https://doi.org/10.1145/3242671.3242680
[12] datdota. 2020. datdota. https://www.datdota.com/ Accessed: November, 2020.
[13]
Thomas Debeauvais, Thomas Zimmermann, Nachiappan Nagappan, Kevin Carter, Ryan Cooper, Dan Greenawalt, and Tyson Solberg. 2014. O with
their assists: An empirical study of driving skill in Forza Motorsport 4. In Proceedings of the 9th International Conference on the Foundations of Digital
17

CHI ’21, May 8–13, 2021, Yokohama, Japan Wallner et al.
Games. 8.
[14]
Anders Drachen and Alessandro Canossa. 2009. Analyzing Spatial User Behavior in Computer Games Using Geographic Information Systems. In
Proceedings of the 13th International MindTrek Conference: Everyday Life in the Ubiquitous Era (Tampere, Finland) (MindTrek ’09). Association for
Computing Machinery, New York, NY, USA, 182–189. https://doi.org/10.1145/1621841.1621875
[15]
Anders Drachen and Matthias Schubert. 2013. Spatial game analytics and visualization. In IEEE Conference on Computational Inteligence in Games.
IEEE, Washington, DC, USA, 1–8.
[16]
Brandon Drenikow and Pejman Mirza-Babaei. 2017. Vixen: Interactive Visualization of Gameplay Experiences. In Proceedings of the 12th International
Conference on the Foundations of Digital Games (Hyannis, Massachusetts) (FDG ’17). Association for Computing Machinery, New York, NY, USA,
3:1–3:10. https://doi.org/10.1145/3102071.3102089
[17]
Meng Du and Xiaoru Yuan. 2020. A survey of competitive sports data visualization and visual analysis. Journal of Visualization (2020), 1–21.
https://doi.org/10.1007/s12650-020- 00687-2
[18] Elo Entertainment. 2020. Dotabu. https://www.dotabu.com/ Accessed: November, 2020.
[19] Ensemble Studios. 1997. Age of Empires. Game [PC]. Xbox Game Studios, Redmond, WA, US.
[20] Epic Games. 2017. Fortnite. Game [PC]. Epic Games, Cary, NC, US.
[21]
William A. Hamilton, Oliver Garretson, and Andruid Kerne. 2014. Streaming on Twitch: Fostering Participatory Communities of Play within Live
Mixed Media. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. Association for Computing Machinery, New York,
NY, USA, 1315—-1324. https://doi.org/10.1145/2556288.2557048
[22]
Erik Hazzard. 2014. Data Visualization in Games. http://erikhazzard.com/blog/game- development/data-visualization- in-games Accessed: April,
2020.
[23]
Nate Hoobler, Greg Humphreys, and Maneesh Agrawala. 2004. Visualizing competitive behaviors in multi-user virtual environments. In IEEE
Visualization. IEEE, Washington, DC, USA, 163–170.
[24] Michael Karlsson. 2016. Information Visualisation in Games: How do you know what you know? Master’s thesis. Umea University.
[25] Andrew Karpushin and Clement Durand. 2020. WOTInspector. https://wotinspector.com/en/ Accessed: August, 2020.
[26]
Athanasios Vasileios Kokkinakis, Simon Demediuk, Isabelle Nölle, Oluseyi Olarewaju, Sagarika Patra, Justus Robertson, Peter York, Alan Pedrassoli
Pedrassoli Chitayat, Alistair Coates, Daniel Slawson, Peter Hughes, Nicolas Hardie, Ben Kirman, Jonathan Hook, Anders Drachen, Marian F Ursu,
and Florian Block. 2020. DAX: Data-Driven Audience Experiences in Esports. In ACM International Conference on Interactive Media Experiences.
Association for Computing Machinery, New York, NY, USA, 94–105. https://doi.org/10.1145/3391614.3393659
[27]
Yen-Ting Kuan, Yu-Shuen Wang, and Jung-Hong Chuang. 2017. Visualizing Real-Time Strategy Games: The Example of StarCraft II. In 2017 IEEE
Conference on Visual Analytics Science and Technology (VAST). IEEE, Washington, DC, USA, 71–80. https://doi.org/10.1109/VAST.2017.8585594
[28]
Yun-En Liu, Erik Andersen, Richard Snider, Seth Cooper, and Zoran Popović. 2011. Feature-Based Projections for Eective Playtrace Analysis. In
Proceedings of the 6th International Conference on Foundations of Digital Games. Association for Computing Machinery, New York, NY, USA, 69–76.
https://doi.org/10.1145/2159365.2159375
[29]
Eva Mayr, Nicole Hynek, Saminu Salisu, and Florian Windhager. 2019. Trust in Information Visualization. In EuroVis Workshop on Trustworthy
Visualization (TrustVis), Robert Kosara, Kai Lawonn, Lars Linsen, and Noeska Smit (Eds.). The Eurographics Association, 25–29. https://doi.org/10.
2312/trvis.20191187
[30]
Adam W. Meade and Bartholomew Craig. 2012. Identifying Careless Responses in Survey Data. Psychological Methods 17, 3 (2012), 437–455.
https://doi.org/10.1037/a0028085
[31]
Ben Medler. 2013. Visual Game Analytics. In Game Analytics: Maximizing the Value of Player Data, Magy Seif El-Nasr, Anders Drachen, and
Alessandro Canossa (Eds.). Springer, London, 403–433. https://doi.org/10.1007/978- 1-4471- 4769-5_18
[32]
Ben Medler and Brian Magerko. 2011. Analytics of Play: Using Information Visualization and Gameplay Practices for Visualizing Video Game Data.
Parsons Journal for Information Mapping 3, 1 (2011), 12.
[33]
Truong-Huy D Nguyen, Magy Seif El-Nasr, and Alessandro Canossa. 2015. Glyph: Visualization Tool for Understanding Problem Solving Strategies
in Puzzle Games. In Proceedings of the 10th International Conference on the Foundations of Digital Games. 9.
[34] OP.GG. 2020. PUBG Stats, Matches, Leaderboards, Guides, Weapons - PUBG.OP.GG. https://pubg.op.gg/weapons Accessed: August, 2020.
[35] P UBG Corporation. 2017. PUBG. Game [PC]. PUBG Corporation, Bundang-gu, South Korea.
[36] Riot Games. 2009. League of Legends. Game [PC]. Riot Games, Santa Monica, CA, US.
[37]
Hans-Jörg Schulz, Thomas Nocke, Magnus Heitzler, and Heidrun Schumann. 2013. A Design Space of Visualization Tasks. IEEE Transactions on
Visualization and Computer Graphics 19, 12 (2013), 2366–2375.
[38]
Magy Seif El-Nasr, André Gagné, Dinara Moura, and Bardia Aghabeigi. 2013. Visual Analytics Tools – A Lens into Player’s Temporal Progression
and Behavior. In Game Analytics: Maximizing the Value of Player Data, Magy Seif El-Nasr, Anders Drachen, and Alessandro Canossa (Eds.). Springer,
London, 435–470. https://doi.org/10.1007/978-1- 4471-4769-5_19
[39] Shadow Esports GmbH. 2020. Shadow – Game-changing esports analytics. https://shadow.gg/ Accessed: November, 2020.
[40]
Ben Shneiderman. 1996. The eyes have it: a task by data type taxonomy for information visualizations. In Proceedings 1996 IEEE Symposium on
Visual Languages. IEEE, Washington, DC, USA, 336–343. https://doi.org/10.1109/VL.1996.545307
[41]
Max Sjöblom and Juho Hamari. 2017. Why do people watch others play video games? An empirical study on the motivations of Twitch users.
Computers in Human Behavior 75 (2017), 985–996. https://doi.org/10.1016/j.chb.2016.10.019
18

What Players Want: Information Needs of Players on Post-Game Visualizations CHI ’21, May 8–13, 2021, Yokohama, Japan
[42] Skybox. 2020. Skybox. https://skybox.gg/early- access Accessed: November, 2020.
[43] The OpenDota Project. 2020. OpenDota. https://www.opendota.com/ Accessed: August, 2020.
[44] Valve. 2013. Dota 2. Game [PC]. Valve, Bellevue, WA, US.
[45]
Günter Wallner. 2018. Automatic generation of battle maps from replay data. Information Visualization 17, 3 (2018), 239–256. https://doi.org/10.
1177/1473871617713338
[46]
Günter Wallner. 2019. A Brief Overview of Data Mining and Analytics in Games. In Data Analytics Applications in Gaming and Entertainment,
Günter Wallner (Ed.). CRC Press, Boca Raton, FL, US, 1–14.
[47]
Günter Wallner, Alessandro Canossa, and Magy Seif El-Nasr. 2018. Introduction to the special issue on visual game analytics. Information
Visualization 17, 3 (2018), 181–182. https://doi.org/10.1177/1473871617722040
[48]
Günter Wallner, Nour Halabi, and Pejman Mirza-Babaei. 2019. Aggregated Visualization of Playtesting Data. In Proceedings of the 2019 CHI
Conference on Human Factors in Computing Systems. Association for Computing Machinery, New York, NY, USA, Article 363, 12 pages. https:
//doi.org/10.1145/3290605.3300593
[49]
Günter Wallner and Simone Kriglstein. 2013. Visualization-based analysis of gameplay data – A review of literature. Entertainment Computing 4, 3
(2013), 143–155. https://doi.org/10.1016/j.entcom.2013.02.002
[50]
Günter Wallner and Simone Kriglstein. 2016. Visualizations for Retrospective Analysis of Battles in Team-Based Combat Games: A User Study. In
Proceedings of the 2016 Annual Symposium on Computer-Human Interaction in Play. Association for Computing Machinery, New York, NY, USA,
22–32. https://doi.org/10.1145/2967934.2968093
[51]
Matthew O. Ward, Georges Grinstein, and Daniel Keim. 2010. Interactive Data Visualization: Foundations, Techniques, and Applications (1 ed.). CRC
Press, Boca Raton, FL, US.
[52] Wargaming Minsk. 2010. World of Tanks. Game [PC]. Wargaming, Nicosia, Cyprus.
[53] Westwood Studios. 1995. Command & Conquer. Game [PC]. Virgin Interactive Entertainment, London, UK.
[54] Veronica Zammitto. 2008. Visualization Techniques in Video Games. In Electronic Visualisation and the Arts (EVA 2008). BCS, 267–276.
19