How can online games retain players?

For an online game to succeed, it needs to retain customers.  That’s a simple and obvious truth.  Less simple and obvious though are the methods by which this can be accomplished.

I see four critical design components to retaining customers.  Two more – customer service and appropriate pricing for value – aren’t design components are thus are not in the scope of the discussion here.  I’m going to focus on ongoing novelty, concrete goals, playstyle support, and social ties.

Ongoing Novelty

Ongoing novelty is, simply, the idea that new experiences are continually available to players.  When players feel that they have experienced all the content in all the meaningful ways they can do so, they’ll usually lose interest pretty quickly.  It’s thus important to have new experiences available at all times for players.  This can be accomplished in two main ways: by gating content, and by having content provide variable experiences.  Let’s look at each in turn.

Gated content is content that isn’t available to players until they’ve accomplished some task in game.  Tasks include things like reaching a certain level, acquiring certain gear, or completing certain quests.  With the prerequisite tasks out of the way, players have more content to look forward to.  Gated content is very popular and exists in almost all online games, but it has a serious drawback – once players have completed the prerequisites and reached the gated content, they will often complete it then lose interest, and always look for other, novel content to experience.

Variable experiences can in turn be achieved in two basic ways: by varying the content itself, or by relying on player variations.  Varying the content itself would entail things like randomizing terrain, encounters, or loot tables, such that players would want to experience the “same” content repeatedly because it won’t truly be the same each time.  This approach certainly works, to a certain point anyhow… but the degree of perceived novelty still drops over time and players will eventually lose interest in repeating the same experience with minor changes.  The greater the changes in the experience, the more effective this approach is at retaining players’ interest.

Relying on player variation is a key idea and one that’s rarely fully developed.  Fundamentally, players are a lot more variable than game software, and so they provide a far greater (and renewable!) source of variation for games.  This is one of the greatest reasons why PvP is popular with many players; it’s a different experience each time because people are unpredictable.  To a lesser extent, raiding also relies on player variation, in that the success of a group is dependent on the skills and attitudes of the players involved; if you attempt the same raid twice with two different groups, it won’t be the same experience.  There’s a further (and very important) element to player variation that I’ll address under the section on Social Ties below.

Concrete Goals

People play games for a wide variety of reasons, and it’s unwise to believe that everyone has the same goals.  It’s however true that many players like to have personal goals when they play games, and strive to work towards these goals.  To the extent that an online game allows a variety of possible goals, and supports progress towards said goals, that game will better retain its playerbase.  Let’s look at some examples.

First and most obvious is the very common goal of obtaining increasing power within a game.  Power can be represented by levels, gear, other types of wealth, reputation, skills, political power, etc.  Notably, the first five of those six means of representing power in game are concretely supported by game mechanics, and are easy to measure and compare.  Players can in most cases readily establish who has more money/better gear/higher levels by comparing stats.  Moreover, power measured in these ways is recognized by greater freedom within the game; e.g. a wealthy player is free to buy more goods and services, and a higher level character is free to travel in more places safely.  Power can also be represented by freedom to explore gated content, and thus can be linked to completion of other content – so in a sense a player’s power increases with every quest completed, as those completed quests may unlock new quests or areas.  Political power is actually an emergent goal so will be defined below, and examined more closely in the section on Social Ties.

A second and obvious way games can provide goals is through achievement systems that track player activity.  Achievements do not necessarily provide any power to the player in mechanical in-game terms, but can provide satisfaction nonetheless.  Achievements can track activities like exploring areas, defeating monsters, overcoming other challenges, etc.  Sometimes achievements yield rewards that are not mechanically related to the game but are still of interest to players, e.g. non-combat pets or titles.

Less obvious but still important are emergent goals; these are goals that players create for themselves that aren’t explicitly part of the game’s rules.  One of the more common types of emergent goals is intended to increase difficulty of content by creating artificial limitations.  E.g. a player might decide to attempt to win a fight wearing no armour, or a group of players might attempt to win a fight with fewer players than are allowed by the game’s rules.  Other types of emergent goals might be to e.g. own all the pets in the game, or dress one’s character in attractive clothing.  Some emergent goals are social, e.g. wanting to make friends in-game, or wanting to exert political influence in-game.  I’ll discuss both of these a bit under Social Ties below.  Emergent goals are quite common in my experience, and they all rely to some extent on the amount of flexibility in the mechanics and the breadth of content available.  Emergent goals are also often limited by the amount of agency available to players, which is discussed in Social Ties below.

Goals that are vague and ill-defined don’t tend to be as popular as well-defined, concrete goals – whether we’re talking about mechanically supported goals or emergent goals.  Players like knowing that they’re able to measure their progress towards goals, and to feel a sense of accomplishment thereby.  If e.g. one of my personal (power) goals is to master a certain craft, and one of my personal (emergent) goals is to create one of every item that can be created with that craft, then I’ll be happiest if I have ways to know that my time spent in game is increasing my progress towards those goals.  If e.g. the game has a completely random crafting advancement system, where crafting skill can increase or decrease randomly, and no feedback is given to the player about current crafting skill, how can I ever be sure that I’m getting better overall?  How can I meaningfully pursue that goal?  Though my goal is in this case quite concrete, if the game doesn’t give me a means to measure progress towards that goal, I’m likely to lose interest in that goal.

Playstyle Support

Playstyle Support is the vaguest of these design considerations because it depends the most heavily on the initial definitions of which playstyles are to be supported. Playstyle is the way a player approaches content at any given time, and there are a wide variety of playstyles.  Some of the more common playstyles are defined by Bartle’s agèd but still meaningful definition of player types: Achiever, Explorer, Socializer, and Killer.  Achievers want to have things to achieve; explorers want to have things to explore, etc.  Designers can intentionally (or unintentionally) bias a game towards one or more of these groups, and it’s wise to think about how you’d like to support players enjoying the game.  It’s very difficult to make a game that appeals to everyone, and most games focus on one or two of the above; that’s not to say that games cannot and do not appeal to all four of the above – just that it’s less common to have intentional design considerations that appeal equally to all of Bartle’s player types.

Other ways of looking at playstyle can include defining player preferences as “hardcore” or “casual”, or as “roleplayers” or “non-roleplayers”, or as “groupers” or “soloers”, or as “player vs environment” (PvE) or “player vs player” (PvP).  These terms are notoriously vague and are used in different ways by different people, but that doesn’t mean that they cannot be helpful ways for designers to look at playstyles.  A game that’s intended to cater to casual players might (depending on how “casual” is defined, of course) have content that’s accessible in short periods of time.  A game that’s intended to cater to roleplayers might focus more on game lore and on immersion in the game’s virtual reality than a game that’s intended to cater to non-roleplayers.

In short, designers ought to consider which playstyles they want to explicitly support, and make the game as accessible as possible for those playstyles.  Including elements that aren’t well supported can be very frustrating to players and cause them to lose interest quickly – e.g. when a game designed for PvE content has a token PvP element that doesn’t mesh well with the rest of the game, that’s a problem.  PvEers won’t necessarily care about the PvP content, and PvPers might be unhappy that their playstyle received a very cursory implementation.

Social Ties

Social ties are perhaps the strongest single element in player retention in online games, and yet are often woefully underdeveloped.  People are by nature social creatures, and most of us like to either cooperate or compete with others, whether actively or passively.  That’s indeed one of the main draws in online games – even solo players have a social context in which to measure their achievements (competitive) or can contribute to a collective goal (cooperative).  I believe that there are three essential elements to fostering social ties in online games: communication, grouping, and agency.

Communication is a simple concept yet I’ve seen so many games fail to adequately implement this basic necessity that I’m often quite flummoxed.  Communication most commonly comes in two forms in modern games: voice chat and text chat.  Voice chat is quite popular with a lot of people, and is the preferred method of communicating in some contexts, yet many games still fail to support it.  If voice chat is included, designers should consider issues such as how voice chat is activated (push to chat or always on?); how many channels are available to players (e.g. group, guild, local, tell) and how those are differentiated; how well does voice chat mesh with the game’s sounds; how can players mute unwanted signals, etc.  Few designers seem to ponder all of the above though, and when voice chat is implemented, it’s often riddled with lack of planning that causes problems or leaves players unhappy.

Text chat is hardly a novel concept; it’s the basic and underlying form of communication in most online games, dating back to the times when online games were text-only.  It’s surprising then to see how many games fail to adequately support text chat.  Basic design considerations here include creation of custom channels, ability to readily differentiate between multiple channels of input (e.g. different color text for local as opposed to group), and existence of special and global channels.  In general, greater flexibility is desirable, but it’s good to consider the social ramifications of creating pre-existing text channels, e.g. regional or global chat.  Barrens chat in WoW was frankly a pretty awful experience for many of us, while other players loved it and continue to spread the gospel of Chuck Norris jokes and random bigoted comments.  It’s ok to provide such channels, but it’s wise to think it through beforehand, and to use text channel design to steer the community into the types of interactions you think will best benefit the game.

Grouping is a fundamental idea in online games, and comes in four flavors: short-term groups, long-term groups, exclusionary groups, and contextual groups.

Short-term groups are generally known as parties or groups or fellowships or something of the sort.  These groups exist to allow players to take on challenges that are too difficult for one person to overcome; many games feature content explicitly intended for short-term groups.  To the extent that a game relies on or features such content, it’s good to build in support for finding and forming such groups quickly and easily, as well as tools that allow group members to locate each other and even travel quickly to each other.  Some games benefit from allowing groups to assign leaders (often this is by default the person who formed the group) and further allow the leader to assign roles to team members, often by visually marking them.  One of the dividing issues in groups is how loot is divided, and it’s good to consider how you plan to support loot division and build in a variety of options to support various player preferences.  Random number generators that are accessible to the players in-game are helpful in this regard, as are explicit “loot rules” that the leader can assign for the party, e.g. Master Looter (i.e. leader takes all and divides it later), Round Robin (i.e. each player may loot in turn), or Free for All (i.e. anything goes).  Some games provide content designed for groups that are larger than default groups (these larger groups are often called “raids”), and to the extent that a game offers such content, it’s wise to support features for these larger groups – such as communication channels, tools for leaders, etc.

Long-term groups include such structures as guilds/kinships, friend lists, and also less defined groups that work for social, political, and economic goals.

Guilds are generally formed either to work collectively towards a common goal or goals, to provide a ready source of short-term groups, or to gather together friends for social activities and chat.  As is common for most player preferences, people who like one type of guild often fail to recognize the existence or validity of the other types of guilds, so there’s a continual disagreement about what guilds do, or what their purpose is.  Regardless, guilds are important to a lot of players, and it’s good for game designers to consider carefully to what extent their game can and should support guilds.  Guilds benefit from game mechanics that give power explicitly to guilds as a group or to guild members (e.g. guild halls with extra storage, instant transportation to guild halls, special quests or instances for guilds), as well as from game mechanics that better enable communication and coordination between guild members.  Allowing guilds to create custom ranks for members, then set permissions for various mechanically supported activities based on rank (e.g. access areas of the guild hall, access storage), increases enjoyment for members of those guilds and foments social ties.

A larger form of long-term group is a group of multiple guilds, often called an “alliance”.  Support for alliances could include any and all of the types of support explicitly given to guilds, including chat channels, alliance-controlled areas with limited access and/or storage, travel to alliance-controlled areas, etc.

Another form of long-term group is the personal list of “friends”, players that a given player enjoys interacting with in game.  A “friends list” is a powerful tool for supporting social interactions between players, and is especially useful for connecting players who are not in guilds or belong to different guilds.  Friends lists often provide a player with information about other players on their lists, including their location, level, and status (e.g. grouped or solo).  One of the potential problems with this concept is that friends lists are locally controlled by the player who “owns” a given list, and so it’s possible for people to use these tools to gather information about other players against their will.  One simple workaround is to allow players to set levels of access for their personal information – e.g. “nobody can read my info”, “people on my friends list can see my info”, and “everyone can see my info”.

Exclusionary groups are formed of people who are excluded by a player from his or her gaming experience, to the extent that the game mechanics support this.  The simplest version of an exclusionary group is the “ignore list”, which generally blocks a player from viewing communications from ignored players.  Exclusionary groups can also be automatically formed by the game software to e.g. block gold spammers.  Sometimes exclusionary groups can exist by default based on player preferences – e.g. blocking all tells that do not come from friends/guild/group members.  Some level of support for exclusionary groups is always desirable, as players can lose interest in games where they are subjected to repeated unwanted social interactions.

Contextual groups are more complex than the other group types above, and are based on social, economic, or political activities.  All of these activities are made available or unavailable based on the level of player agency in the game, so they’ll be discussed further below.

Agency is the amount of control that players can exert on the gameworld.  Not all online games require or would benefit from agency, but many do.  The greater the amount of agency players have, the greater potential there is for emergent behaviour and emergent goals, as noted above.  Some of the more common emergent behaviours that players engage in as groups include social, economic, and political activities, and all of these are defined and limited by the degree of agency the game provides.

Social agency comes from the ability to interact with other players, either through communication, or through mechanical means (e.g. attacking, healing, buffing, debuffing).  If e.g. a game allows players to heal or buff strangers, then agency exists through which the player culture might develop a pricing structure for such services, or people might perform them as acts of charity.  If e.g. a game allows players to perform music in-game, this can lead to social events like concerts, or to musicians frequenting places where players gather.  It’s always good to consider how you can both extend and limit social agency though – e.g. not everyone wants to listen to player-generated music all the time.

Economic agency comes from the ability to create goods, and to trade goods and services.  Crafting skills allow the former, while the latter is supported by in-game mechanics such as trading screens (where both parties can see the goods being offered), auction houses, etc.  It’s wise not to underestimate the amount of interest many players have in economic agency; greater support in this regard can be a powerful incentive to retain players.

Political agency is more complex then social and economic agency because it extends to a wide range of activities.  Politics can involve such issues as status in a guild or other group, and also status with or control over in-game groups or resources.  The latter is an intriguing concept that’s generally woefully underdeveloped, yet promises great potential for ongoing novelty, concrete goals, and social ties.  Control over in-game groups or resources could include things like player control over production resources like mines, social resources like inns or taverns, or simply influence over the policies or actions of groups of NPCs.

Clearly, just as there is great potential here for players to essentially create content for each other, there’s also great potential for players to abuse the game and act in antisocial ways that decrease the enjoyment of other players. In general, the greater the agency that is granted to players, the greater the controls need to be to prevent players from abusing that agency.  In-game controls are always preferable to external controls (e.g. EULAs, Codes of Conduct, etc) because external controls aren’t always understood by players and are far more difficult to enforce.  In-game controls are simple to understand, when well implemented: if the game allows you to do it, it’s ok.  That’s simplistic but it’s how a lot of players approach games so it’s a good starting point for controlling undesirable behaviours.  With that in mind, the more agency players are provided with, the better able the game will be to retain their interest.


4 comments so far

  1. […] posted here:  How can online games retain players? « Fool's Age Civiballs Xmas Edition – Online Flash Games | Online Games 777.comFT Advent Calendar Of Free […]

  2. […] can online games retain players? « Fool's Age no comment Posted by admin wrote an interesting post today onHere’s a quick excerptGated content is very popular and […]

  3. Tesh on

    Wall of text! I love these.

    First and foremost, why retain players in the first place?

    For money? If you’re a sub game, you want to keep chewing on players, and if you’re a microtransaction game, you want to sell stuff (which may or may not strictly require retention), but if you’re Guild Wars, you made your money when you sold the box.

    For game design/playerbase concerns? Well, that’s a different animal. It’s still affected by the business model, though. Assuming this is purely about game design, though, let’s see… in order…

    Content gating also has a serious other problem; it aids and abets evil grindy mechanics. Player variation over content variation is a great direction to go for novelty, though. Players do the weirdest things…

    Regarding concrete measurement, you *can* go too far. I do like some concrete goals to aim for, but when everything is boiled down to incremental progress and Achievements, we’re too close to Progress Quest for me to really enjoy it. I don’t mind if there are lots of concrete coals for those who like them, but I like some abstract freeform content as well for when I don’t want to be constantly measured.

    Playstyle support is key to a smart project, methinketh. Choose a core focus for your game and polish it to a shine. Trying to be everything for everyone just isn’t practical. (This also ties into project scope and management issues, which is probably a tangent too far, though relevant.)

    I’m a bit underqualified to wax eloquent on the social side, but what you have looks good. I’ll heartily second the notion of agency, though. I’ve long argued that greater agency is key to a greater game. To me, that’s the heart of playing a game in the first place. If I wanted purely to sit back and be railroaded through my entertainment, I’d watch a movie.

  4. foolsage on

    Thanks for the thoughtful reply as always, Tesh.

    First, I do assume that online games want to retain players. Either they’re monetized such that more players leads to more money, or they benefit from having a “critical mass” of players available to form a community.

    I agree that content gating can be pretty dubious; it’s most often a carrot used to lure players onto treadmills, and I don’t appreciate that. That’s more a problem though with how the industry uses gated content than a problem with gating content in and of itself. That’s a bit like the NRA’s argument about how guns don’t kill people, but there you are. 😉

    I certainly don’t advocate having every possible goal in game be measured and rated; rather, I prefer that emergent goals be possible and that the game be transparent enough to allow players to have a sense of progress towards their goals, whatever they may be. The more emergent the goals, the harder they are to track of course, and I’m certainly in favour of enabling and supporting emergent gameplay as much as possible.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: