Estimating the Value of Multiplayer Modes in Video Games: An

Analysis of Sales, Ratings, and Utilization Rates

Eric Nelson Bailey and Kazunori Miyata

Japan Advanced Institute of Science and Technology, Ishikawa, Japan

Keywords: Video Game Development, Online Multiplayer, Project Scope, Functional Scope, Project Management,

Knowledge Discovery in Databases (KDD).

Abstract: Video game organizations are under pressure from growing development costs and competition from other

sources of entertainment. Making informed project scope decisions is critical to avoiding budgetary and

schedule waste, but little information regarding best development practices is made public, and many

decisions are made, not informed by past data, but by the tacit knowledge of project owners. One way

developers have attempted to improve sales on their games and follow-up content is through the inclusion of

online multiplayer functionality. However, multiplayer functionality is expensive to develop and can

significantly add to the costs and schedule of a game project, particularly if the decision to include it is

made late in the process. This research explores publicly available data to discover the value that

multiplayer functionality provides so that project owners can make a more informed decision earlier in the

development process.

1 INTRODUCTION

Video game development companies often rely on

the tacit knowledge of project owners, such as

directors and producers, to decide critical project

management issues such as scope, scheduling, and

budgetary requirements. However, these tacit

assumptions are not always checked against results

in the market and may not reflect the reality of the

ever-changing game development business.

Mistakes in scope decisions can lead to resource

waste and, in worst cases, project failure.

There is room for improvement with project

management practices in video game development.

(Musil et al., 2010) One way to improve project

management is by incorporating data from both the

developer and the larger industry’s past results to

better inform future decisions.

In this research, we examine one of the most

time and resource intensive functional scope

decisions: whether or not to include a multiplayer

feature within a game. Through analyzing the sales

and critical results of 34,263 applications on Valve

Corporation’s Steam PC game service, we provide

decision makers with insight into the value including

multiplayer can add to a video game project.

2 RESEARCH BACKGROUND

2.1 Video Game Development

Video game development is expensive, with

development costs alone reaching $100 million for

one title. (Bleeker, 2013) “How much game to

make?” is an important question that needs to be

answered early in the development process to avoid

costly rework later in production.

“Managing the project scope is primarily

concerned with defining and controlling what is and

is not included in the project." (Project Management

Institute, 2017, p. 129) Scope can impact quality and

the ability to deliver a game project on schedule.

(Washburn et al., 2016) Scope in video games

includes both functional scope – the unique features

and systems available in the game – and content

scope – the amount of content provided with the

game. In particular, one aspect of functional scope

that has the potential to increase the amount of time

and cost required to develop a video game project is

an online multiplayer feature. Games need to be

built with multiplayer in mind from the beginning,

as the entire game may need to be designed for it,

and later implementation can force developers to

rework much of what has already been made.

Bailey, E. and Miyata, K.

Estimating the Value of Multiplayer Modes in Video Games: An Analysis of Sales, Ratings, and Utilization Rates.

DOI: 10.5220/0006898201550162

In Proceedings of the 10th Inter national Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2018) - Volume 1: KDIR, pages 155-162

ISBN: 978-989-758-330-8

155

Including a multiplayer feature brings many

potential benefits, but it comes with many

drawbacks as well, particularly the expense involved

in developing and maintaining an online mode.

Even when resources go into developing

multiplayer, not all of the features are used by the

majority of players. (Hullet et al., 2011) Within the

game console market, prior research noted that,

because of the need for an active community,

multiplayer can be a valuable addition to the game

towards the end of a console’s life cycle, when there

are enough players to justify it, while a single player

focus can be a stronger strategy earlier in the console

life cycle. (Marchand, 2016)

However, for PC games, the life cycle dynamics

that dominate the console market are not present. Of

the many PC game retail outlets, Valve

Corporation’s digital distribution platform, Steam,

dominates the market with 18.5 million concurrent

users in January 2018. (xPaw and Marlamin, 2018)

There are positives and negatives to including online

multiplayer functionality, so a more data-informed

approach could help product owners decide early in

the development process whether to pursue an online

strategy or to focus resources on the single-player

game.

2.2 Software Project Management

Game development organizations try to balance

getting projects done on time and in budget with

giving their employees the creative freedom they

need to innovate. (Cohendet and Simon, 2007)

However, few game development processes are well

defined, and developers could benefit from

improved project management processes. (Musil,

Winkler and Biffl, 2010)

One aspect of project management in particular,

scope management, is difficult in all software

development (Brooks, 1995; DeMarco and Lister,

2003; Yourdon, 1997), but more so in video game

development because the non-utilitarian, experiential

nature of the product makes a definition of “value”

elusive. (Bilton, 2007, p. xvii; Hirsch, 1972; Lampel,

Lant and Shamsie, 2000) Deciding on an appropriate

scope is critical to the quality of the final product.

(Washburn et al., 2016)

2.3 Knowledge Discovery in Databases

Data is an effective way to check perceptions and

possibly fix misperceptions within an organization.

(Weber, 2005, p. 14) In particular, many

organizations rely on their existing knowledge,

which can be difficult to change and lead to

complacency or crises, especially when it has led to

success in the past. (Leonard-Barton, 1995; Nystrom

and Starbuck, 1984; Senge, 1990)

Intelligence gathering and learning not just from

an organization’s past history but from the past

history and best practices of other organizations is

an essential quality of a learning organization.

(Garvin, 1993) Even if the final decision comes

down to the project owner’s judgment, data is

essential for good decision making.

For this research, data mining of web-based

sources is used to derive insights that could be

helpful for video game businesses. We also

introduce a technique that could be used to

determine multiplayer usage through examining

video game “achievements” on a limited sample.

3 RESEARCH METHODOLOGY

3.1 Research Questions

The principle research question that we are

attempting to answer is whether the inclusion of a

multiplayer feature is worth the development effort.

We hypothesize that the inclusion of multiplayer has

a positive impact on sales because it adds to the

utility of a video game for players; however, we also

hypothesize that the additional demands required for

a positive multiplayer experience will have a

negative impact on user ratings. We also

hypothesize that this impact will not be felt as

strongly in professional critic ratings because critics

are testing the game in a more ideal, pre-release

environment. Finally, we hypothesize that the

multiplayer features of games that focus on single-

player content are not used by the majority of

players.

3.2 Research Scope

The sample used in this research includes games (PC,

Mac, Linux) released on Valve Corporation’s Steam

service from its inception to early 2017. More recent

data was not included because it could contain

games that are still selling or receiving reviews.

The Steam service was chosen for three reasons:

first, it is the most popular online PC game retail and

digital distribution platform; second, much of

Steam’s data is publicly available, so it is more

suitable for large-scale sampling; third, the service is

a fully digital distribution platform for the PC, so it

is not subject to game console lifecycles.

KDIR 2018 - 10th International Conference on Knowledge Discovery and Information Retrieval

156

3.3 Data Sources

The primary source of data used in this research is

data made publicly available for video games

released on Valve Corporation’s Steam service

platform. This data includes aggregated professional

critic scores from Metacritic.com. In addition, data

from Sergey Galyonkin’s SteamSpy and “xPaw” and

“Marlamin’s” SteamDB was employed for

estimating the number of owners for each game.

Steam’s Top Sellers of 2017 list (Valve

Corporation, 2018), which is divided into

“Platinum”, “Gold”, “Silver”, and “Bronze”

categories by gross revenue, and The NPD Group’s

10 Best-Selling Games of the Year lists from 2001

to 2017 (The NPD Group, 2017) were used to find

examples of the best-selling games on both Steam

and video game consoles.

3.4 Data Preparation

The list of all applications on Steam was obtained

from its API, available from the link:

http://api.steampowered.com/ISteamApps/GetAppLi

st/v0001/ (last accessed April 27

, 2018) The unique

AppID identifier was used as a base for collecting

information from Steam Store pages for each game

using Python to generate links of the form:

http://store.steampowered.com/app/(AppID) (last

accessed April 27

, 2018) These store pages

included tagging information that could be used to

identify whether Valve had labelled the game

content “Single-player”, “Multiplayer”, and/or “Co-

op” as well as the Metacritic score and Steam user

rating – both as percentages. The same AppID was

used with SteamDB, using a link of the form:

https://steamdb.info/app/(AppID) to gather

SteamSpy owner data for the games in the sample.

Many of the applications with AppIDs were not

game applications and needed to be remove from the

data set. This was done by only including entries

that had either a “Single-player”, “Multiplayer”, or

“Co-op” flag. Also, games without a release date

were removed from the sample as they could

indicate unreleased games or incomplete data. This

reduced the sample from 34,263 to 19,621 (57% of

the total). Games with either a “Multiplayer” or a

“Co-op” flag were combined into one multiplayer

category.

3.5 Data Modeling

Means (M) with 95% confidence intervals (95% CI),

medians (Mdn) with interquartile ranges (IQR), and

maximums were calculated for owner counts,

Metacritic scores, and user ratings of the total

sample as well as for each category.

Two categories were created for sorting

Metacritic scores and Steam user ratings based on

Metacritic’s game sorting: Favorable (>= 75%) and

Unfavorable (< 75%). (Metacritic, 2018) Statistics

for each category were also calculated and analyzed

for possible differences in trends between

unfavorably and favorably rated games.

The data could not be expected to be normally

distributed, so to determine whether the number of

owners, Metacritic scores, and user ratings for

single-player, multiplayer, single-player only, and

multiplayer only games differed significantly from

those of the overall sample population, a Wilcoxon

rank-sum test was performed on each binary factor

because it does not assume a normal distribution.

3.6 Outlier Testing

To discover outliers for further analysis, examples at

the extremes for each of the categories were

examined. In addition, The NPD Group’s 10 Best-

Selling Game lists for the period from 2001 – 2017

and Steam’s Top Sellers of 2017 list were used to

find examples at the top-selling end of the video

game spectrum.

3.7 Achievement Analysis

As the results in Hullet et al. (2011) suggest,

different aspects of multiplayer might be more

utilized by players than others. In prior research,

achievement data was used to analyze the utilization

rates of single-player content. (Bailey and Miyata,

2017) For this research, Steam achievements were

analyzed for single-player games with multiplayer

modes from the “Platinum”, “Gold”, and “Silver”

categories of Steam’s Top Sellers of 2017 list to

discover trends in player utilization of multiplayer

functionality and modes in games with single-player

content. Not all games have achievements related to

multiplayer usage, but several cases were found in

Steam’s list.

4 RESULTS

4.1 Multiplayer Functionality

Of the 19,621 games in the sample, 18,220 games

(93%) were tagged by Steam as “Single-player”,

Estimating the Value of Multiplayer Modes in Video Games: An Analysis of Sales, Ratings, and Utilization Rates

157

6,312 games (35%) were “Multiplayer”, and 5,071

games (26%) were “Co-op”. “Multiplayer” and “Co-

Op” games were combined for a total of 8,285

games (42%) with multiplayer functionality. Games

can be tagged in more than one category; only

11,336 games (58%) were single-player only and

1,401 games (7%) were multiplayer only.

Steam’s Top Sellers of 2017 list is divided into

four tiers based on gross revenue: “Platinum”,

“Gold”, “Silver”, and “Bronze”. In the Platinum tier,

only one game, The Witcher III, was single-player

only. However, single-player only games represent

roughly a third of games in the remaining tiers. The

distribution of game types is listed in Table 1.

Table 1: The number of games in each category of Steam’s

Top Sellers of 2017.

Category Single-

Player Only

Multiplayer

Only

Both

Platinum

(n = 12)

(8%)

(33%)

(58%)

Gold

(n = 12)

(33%)

(8%)

(58%)

Silver

(n = 14)

(36%)

(29%)

Bronze

(n = 56)

(32%)

(7%)

(61%)

All

(n = 94)

(30%)

(15%)

(55%)

4.2 Multiplayer and Sales

The first relationship tested was between the

presence of single and multiplayer functionality and

sales. (Table 2) This was done by examining the

Steam Spy estimated owner counts for entries with

the three flags. 4,943 samples had Steam Spy owner

counts. The most owned multiplayer game in the

Table 2: The average number of owners for single and

multiplayer games on Steam (in thousands).

Type M

(95% CI)

Mdn

(IQR)

Single Player

(n = 4,816)

146K

(132 – 160K)

23K

(7 – 104K)

Single Only

(n = 3,508)

117K

(104 – 130K)

21K

(6 – 89K)

Multiplayer

(n = 1,435)

267K

(206 – 329K)

32K

(9 – 167K)

Multi Only

(n = 127)

722K

(157 – 1,286K)

24K

(5 – 186K)

Single & Multi

(n = 1,308)

223K

(185 – 262K)

33K

(9 – 163K)

All

(n = 4,943)

161K

(141 – 181K)

24K

(7 – 106K)

sample was Counter-Strike: Global Offensive with

30.6 million owners, while the most owned single-

player game in the sample was Half-Life 2 with 9.9

million owners.

The mean number of owners for multiplayer (M

= 267K) and multiplayer-only (M = 722K) games

was higher than the overall sample (M = 161K) and

much higher than single-player only games (M =

117K). However, median sales showed a different

trend, with games having both single and

multiplayer modes (Mdn = 32K) being higher than

the overall sample (Mdn = 24K). The presence of a

few hit multiplayer-only titles likely pushed the

mean high, but the median owners for multiplayer-

only games (Mdn = 24K) was similar to that of the

overall sample (Mdn = 24K).

Across the categories, a low median compared to

the mean indicates a strong right skew, which could

be expected given the “hit” nature of the video game

business – a few highly owned outlier titles pull the

mean higher.

A Wilcoxon rank-sum test on the presence of

multiplayer showed a statistically significant

difference (p < .001). Subdividing the data into

“Favorable” and “Unfavorable” user rating and

Metacritic score categories did not affect the results.

4.3 Multiplayer and Ratings

The next relationship tested was between the

presence of single and multiplayer functionality and

Metacritic scores (Table 3). Only a subset (n =

2,506) of games have enough professional reviews

to have a Metacritic score.

Table 3: The average Metacritic scores for single and

multiplayer games on Steam.

Type M

(95% CI)

Mdn

(IQR)

Single

(n = 2,417)

72%

(72 – 73%)

74%

(66% - 80%)

Single Only

(n = 1,449)

72%

(71 – 72%)

73%

(66% - 79%)

Multiplayer

(n = 1,057)

73%

(72 – 74%)

75%

(66 – 81%)

Multi Only

(n = 89)

73%

(70 – 76%)

76%

(67 – 82%)

Single & Multi

(n = 968)

73%

(72 – 74%)

75%

(66 – 80%)

All

(n = 2,506)

72%

(72 - 73%)

74%

(66 – 80%)

The mean Metacritic scores are similar across

categories, but the median was slightly higher for

games with multiplayer (Mdn = 75%) and

KDIR 2018 - 10th International Conference on Knowledge Discovery and Information Retrieval

158

multiplayer-only (Mdn = 76%) versus the overall

sample (Mdn = 74%). A Wilcoxon rank-sum test on

the presence of multiplayer on Metacritic scores

indicated that its presence did result in a statistically

significant difference (p < .001) from the overall

sample.

After subdividing the data into “Favorable” and

“Unfavorable” Metacritic score categories, the effect

of multiplayer presence on Metacritic scores was no

longer statistically significant for Unfavorable

(Metacritic score < 75%) games.

The third relationship tested was between the

presence of single and multiplayer functionality and

Steam user ratings. Given that users can rate all of

the games on Steam, the sample available for testing

user ratings is larger than the critic sample. Of the

19,621 games in the sample, 16,189 have user

ratings associated with them. However, a majority of

games have only a few reviews attached to them. To

avoid overweighing games with few reviews, only

games with more than 30 reviews were used for

testing. This left 7,893 samples for testing. The

results are shown in Table 4.

Table 4: The average user review scores for single and

multiplayer games on Steam.

Type M

(95% CI)

Mdn

(IQR)

Single

(n = 7,470)

75%

(74 – 75%)

79%

(65 – 89%)

Single Only

(n = 4,914)

75%

(74 – 75%)

79%

(65 – 89%)

Multiplayer

(n = 2,979)

73%

(72 – 73%)

76%

(62 – 87%)

Multi Only

(n = 423)

65%

(63 – 66%)

66%

(53 – 80%)

Single & Multi

(n = 2,556)

74%

(73 – 75%)

78%

(64 – 88%)

All

(n = 7,893)

74%

(74%)

78%

(64 – 88%)

The user rating was slightly lower for

multiplayer games (M = 73%, Mdn = 76%) and

much lower for multiplayer-only games (M = 65%,

Mdn = 66%) versus other categories. To test whether

this effect held for popular games with more owners,

the means and medians were checked for games

with more than 100K owners. While the average

user ratings rose (n = 464, M = 74%, Mdn = 78%)

for multiplayer games, the average also rose for the

overall sample (n = 1,215, M = 77%, Mdn = 81%),

and the difference remained.

A Wilcoxon rank-sum test on the single-player

flag indicated that the presence of single-player

resulted in a statistically significant difference (p

< .001) in user ratings from the overall sample. The

presence of multiplayer also resulted in a statistically

significant (p < .001) difference. However, when the

data was divided into “Favorable” (user rating >=

75%) and “Unfavorable” (user rating < 75%) groups,

there was no longer a statistically significant

difference for the presence of multiplayer in

“Unfavorable” games.

4.4 Multiplayer Usage

Although it is not possible to make statistically

significant judgments based on a limited sample, we

examined a few cases of how many players took

advantage of multiplayer features when they were

included in single-player games. The achievement

data for games with multiplayer modes in the

Platinum, Gold, and Silver tiers of Steam’s Top

Sellers of 2017 list was inspected for indications of

multiplayer feature usage in games with single-

player content.

In the Platinum tier, Divinity: Original Sin 2

contained a co-operative play mode that could

optionally be used throughout the game’s campaign,

but it also released two special modes for use with

other players. One mode, “Game Master”, allowed a

player to design and host a game for others. The

“Gather Your Party” achievement showed that only

4.2% of players tried the Game Master mode. The

other mode, Arena, gave players a chance to fight

each other. The achievement “Venture Forth”

showed that only 3.5% of players began a match in

this mode.

Within the Gold tier, the game Total War:

Warhammer has both single and multiplayer modes,

but the achievement “Up For A Scrap” showed only

27.5% of players started a multiplayer battle, with

“Armchair General” indicating 18.7% continued to

play for at least five battles and “Armchair Emperor”

indicating that 7.5% continued to play for 25 battles.

Total War: Warhammer also had a longer

multiplayer campaign mode. “A Confrontational

Nature” showed 23.1% of players participated in a

multiplayer campaign, while “First Among Equals”

indicates only 2.9% went on to win it. Within the

Silver tier, the sequel, Total War: Warhammer II

demonstrated a similar pattern of use. “Spoiling for a

Fight” indicated 26.3% of players participated in a

multiplayer battle, while “First Among Equals”

indicated 10.9% remained for 10 multiplayer battles.

“A Taste for Glory” showed 22.4% participated in a

multiplayer campaign.

Another game in the Gold tier, Civilization VI,

also has both single and multiplayer modes. The

Estimating the Value of Multiplayer Modes in Video Games: An Analysis of Sales, Ratings, and Utilization Rates

159

“Land Party” achievement showed that 28.6% of

players participated in at least one multiplayer game.

In the Silver tier game XCOM 2, “The Most

Dangerous Game” indicated only 2.0% of players

went on to win a multiplayer match. In another

Silver tier game, Path of Exile, players could join up

with up to six others. “Band Together” showed

13.6% of players joined a party. However, another

game mode, “Capture the Flag”, was less popular;

only 0.8% of players captured a flag according to the

statistics for the “Capture the Flag” achievement.

4.5 Single-Player Outliers

Although there may be a perceived pressure that

online multiplayer is required in the current market

to succeed, there are examples of single-player only

games that have sold well. Three examples of

console games without online multiplayer features

from The NPD Group’s 10 Best-Selling Games of

2017 list were: Nintendo’s Super Mario Odyssey,

The Legend of Zelda: Breath of the Wild, and

Ubisoft’s Assassin’s Creed: Origins. In the latter

case, previous installments of the Assassin’s Creed

series featured online multiplayer, so it is notable

that it was not included in the most recent game.

Also, The Witcher III made Steam’s Top Sellers

of 2017 highest “Platinum” category despite having

only a single-player experience. Although The

Witcher III was the only example in the “Platinum”

gross revenue list, roughly a third of games in the

“Gold” (4 out of 12), “Silver” (5 out of 14), and

“Bronze” (18 out of 56) were single-player only

experiences.

5 DISCUSSION

5.1 Multiplayer Effect on Sales

The data indicates that games including an online

multiplayer feature tend to do better in sales than

games without one. As one goal in implementing a

multiplayer feature is to create a hit, it is important

to consider both the mean, even if it is sensitive to

outliers, and the median values. In particular, the

“owners” ceiling is much higher for the best-selling

outliers, which results in a mean of 117K owners for

single-player only games versus a higher mean of

223K owners for games with a single-player mode

and a multiplayer mode. The median values are

closer, with 21K for single-player only games versus

33K for games with both modes.

Also, although there are some notable exceptions,

most of the games in Steam’s Top Sellers of 2017

list and The NPD Group’s 10 Best-Selling Games of

2017 list do have online multiplayer. However, even

if games with multiplayer tend to sell better, the

sales numbers may not justify the development risks

and expenses involved.

5.2 Single-Player Only Option

One important observation to make is that every

example of a best-selling game, in both The NPD

Group’s top 10 best-selling games of 2017 list and

the “Platinum” and “Gold” categories of Steam’s

Top Sellers of 2017 list, without online multiplayer

contained a significant volume of single-player

content. For developers aiming for a top-selling hit,

the question comes down to whether it is better to

devote resources to expanding functionality by

adding online multiplayer, and risk diluting the

single player experience, or by expanding or

improving the single-player content. However,

adding single-player content is expensive, and may

only add value for a limited number of players.

(Bailey and Miyata, 2017) Cuphead and Resident

Evil VII, two examples from the “Silver” category,

do show that games with a more compact single-

player experience still have the potential to sell well.

However, the lack of shorter, more linear single-

player experiences from the top of both lists does

seem to indicate that it is unlikely for such a game to

reach the top. A developer either has to focus on

creating a strong multiplayer experience or an open,

expansive single-player experience – both of which

are development resource-intensive, high-risk, high-

reward options.

From a ratings perspective, based on Metacritic

scores, there is no significant advantage to critical

reception for including an online multiplayer feature.

However, when it comes to user ratings, the data

showed that games including online multiplayer

tended to fare worse than single-player games. This is

likely due to user reactions to a range of issues, such

as player and connectivity issues, and inspection of

negative user reviews for games with multiplayer

features supports this could be the case. An absent

online community could certainly impact the multi-

player experience, but the data showed that popularity

is not the only issue; even for games with more than

100K owners, user ratings were lower for games with

multiplayer than games in the overall sample. This

seems to indicate that the inclusion of multiplayer will

tend to result in lower user ratings, possibly due to

issues outside of the developer’s control.

KDIR 2018 - 10th International Conference on Knowledge Discovery and Information Retrieval

160

Some games with a single-player focus but some

multiplayer features did have a significant portion of

players trying the multiplayer content. Total War:

Warhammer I and II as well as Civilization VI had

more than a quarter of its user base trying

multiplayer. However, only roughly a tenth of

players continued to use the multiplayer features in

Total War: Warhammer II for at least ten matches,

indicating the fall-off could be steep in some cases.

In games with strong single-player content,

multiplayer may not be necessary. In one example

from the limited achievement sample, XCOM 2, only

2% of players won a multiplayer match against other

players online, which indicates few players obtained

utility from the multiplayer functionality.

5.3 Multiplayer Variety

Even if the decision is made to include online

multiplayer, reducing the amount of variety to focus

on more popular modes is an option for controlling

development costs without reducing the utility

provided to players. When analyzing data from

Microsoft’s Xbox 360 racing game Project Gotham

Racing 4, Hullet et al. (2011) found that 30 - 40% of

the content in the features they analyzed was used in

less than 1% of races, 12 of 29 event types were

used less than 1% of the time, and 50 out of 134

unique vehicles were used in less than 0.25% of

races.

One example from this research, Path of Exile’s

“Capture the Flag” mode was used by less than 1%

of players. In Divinity: Original Sin 2, less than 5%

of players used the multiplayer “Game Master”

mode, and less than 4% of players fought each other

online in the “Arena” mode. Both games have a

large number of players, meaning the modes did not

go unused, but they only contributed value for a

fraction of the user base and may not have justified

the development time and cost.

Lower user ratings for multiplayer games suggest

the hurdle to developing satisfying multiplayer

content is higher. Players not only need functioning

multiplayer, they also need enough other players

using the same online features. If a mode or

multiplayer in general is not likely to support a large

enough community, then it would be better left out

of the game.

5.4 Directions for Future Research

Further research into the value and utilization of

multiplayer functionality in games is required to

help project owners make better decisions.

Currently, few numbers are available regarding

development costs for multiplayer projects. What

numbers do exist are either anecdotal, or are given in

a post-mortem fashion, which has two problems:

post-mortems tend to have a success bias because

they are done for games that survive to release, and

post-mortems only cover the final numbers, not the

initial projections or the amount of resources lost

due to scope underestimation or change. Detailed

research into the actual cost and schedule impact of

multiplayer is required, particularly on the impact of

making the decision in the early phases of

development versus later phases.

Also, although a limited sample of multiplayer

usage rates based on Steam achievements was

included, a comprehensive survey of multiplayer

usage rates is needed to provide specific insight into

multiplayer utility value, particularly the multiplayer

value to players from different implementation

varieties.

Finally, text mining Steam Store rating and

community pages for multiplayer issues could yield

valuable insights into which areas in particular

contribute to lower user ratings.

6 CONCLUSION

Developers not only need to make scope decisions

based on the potential value that scope will provide

to their players, developers need to make those

decisions early in the process and stick to them to

avoid the costs that come from re-scoping.

Given that the highest selling games tend to have

multiplayer, and that the average number of owners

with both single and multiplayer features was higher

than that of games with only single-player

functionality, there seems to be a sales benefit to

including multiplayer. However, relatively low

utilization of multiplayer features in games with a

single-player focus seems to indicate that forcing a

multiplayer feature into a game will not provide

additional value to most users, and could result in

lower satisfaction, as judged by user ratings and

reviews. Multiplayer-only games have the potential

to become the biggest hits, and do not appear to

suffer from the lack of single-player features.

However, players have high expectations for

multiplayer; if these expectations cannot be met,

results can suffer.

Traditionally, there has been a pressure to

implement multiplayer features because they add

utility for players, discourage piracy, and prevent

used sales, but in a digital distribution environment,

Estimating the Value of Multiplayer Modes in Video Games: An Analysis of Sales, Ratings, and Utilization Rates

161

the latter two reasons no longer apply. The results of

this research demonstrated that there is value to

including multiplayer functionality; however, that

value may not be enough to justify the additional

costs, time, and risks added to development or the

diluting of the single-player experience. Based on

the data, unless a developer is confident that their

multiplayer mode will be capable of maintaining a

strong audience as well as contribute significantly

and positively to the game experience, it would be

better to avoid it if the costs and risks of

implementing it are high.

REFERENCES

Bailey, E. N., Miyata, K., 2017. Analyzing video game

completion achievements: Implications for game

project scope. 16th International Conference on

Entertainment Computing, 469-472. doi: 10.1007/978-

3-319-66715-7_65

Bilton, C., 2007. Management and creativity: From

creative industries to creative management, Blackwell

Publishing.

Bleeker, E., 2013. Grand Theft Auto 5 sales hit $1 billion,

will outsell entire global music industry [Online].

Available:

https://www.fool.com/investing/general/2013/09/28/gt

a-5-sales-hit-1-billion.aspx [Accessed February 28,

2017].

Brooks, F. P. J., 1995. The mythical man month,

anniversary edition: Essays on software engineering,

Pearson Education.

Cohendet, P., Simon, L., 2007. Playing across the

playground: Paradoxes of knowledge creation in the

videogame firm. Journal of Organizational Behavior,

28(5), 587-605. doi: 10.1002/job.460

DeMarco, T., Lister, T., 2003. Waltzing with bears:

Managing risk on software projects, New York, NY,

Dorset House.

Garvin, D. A., 1993. Building a learning organization.

Harvard Business Review, 71(4), 78-91.

Hirsch, P. M., 1972. Processing fads and fashions: An

organization-set analysis of cultural industry system.

American Journal of Sociology, 77(4), 639-659. doi:

10.1086/225192

Hullet, K., Nagappan, N., Schuh, E., Hopson, J., 2011.

Data analytics for game development. Proceedings of

the 33rd International Conference on Software

Engineering, 2011, ICSE '11, 940-943. doi:

10.1145/1985793.1985952

Lampel, J., Lant, T., Shamsie, J., 2000. Balancing act:

Learning from organizing practices in cultural

industries. Organization Science, 11(3), 263-269. doi:

10.1287/orsc.11.3.263.12503

Leonard-Barton, D., 1995. Wellsprings of knowledge,

Boston, MA, Harvard Business School Press.

Marchand, A., 2016. The power of an installed base to

combat lifecycle decline: The case of video games.

International Journal of Research in Marketing, 33(1),

140-154. doi: 10.1016/j.ijresmar.2015.06.006

Metacritic, 2018. How we create the Metascore magic

[Online]. CBS Interactive, Inc. Available:

http://www.metacritic.com/about-metascores

[Accessed April 30, 2018 2018].

Musil, J., Schweda, A., Winkler, D., Biffl, S., 2010. A

survey on the state of the practice in video game

software development. Technical Report No. IFS-QSE

10/04.

The NPD Group's 10 Best-Selling Game Lists (2001-

2017), 2017. The NPD Group.

Nystrom, P. C., Starbuck, W. H., 1984. To avoid

organizational crises: Unlearn. Organizational

Dynamics, 12(4), 53-65. doi: 10.1016/0090-

2616(84)90011-1

Project Management Institute, 2017. A guide to the project

management body of knowledge (PMBOK guide),

Newtown Square, Pennsylvania, Project Management

Institute, Inc.

Senge, P. M., 1990. The fifth discipline: The art and

science of the learning organization, New York, NY,

Currency Doubleday.

Valve Corporation, 2018. Best of 2017: Top sellers

[Online]. Available:

https://store.steampowered.com/sale/2017_best_sellers

/ [Accessed April 30, 2018 2018].

Washburn, M. J., Sathiyanarayanan, P., Nagappan, M.,

Zimmerman, T., Bird, C., 2016. What went right and

what went wrong: An analysis of 155 postmortems

from game development. Proceedings of the 38th

International Conference on Software Engineering

Companion, 2016, ICSE '16, 280-289. doi:

10.1145/2889160.2889253

Weber, A., 2005. Data-driven business models, South-

Western Pub.

xPaw, Marlamin, 2018. Steam online charts: Concurrent

steam users [Online]. Available:

https://steamdb.info/graph/ [Accessed April 30, 2018

2018].

Yourdon, E., 1997. Death march, Prentice Hall PTR.

KDIR 2018 - 10th International Conference on Knowledge Discovery and Information Retrieval

162