I am generally reluctant to put into the public space any detailed analysis of existing products, for fear that someone's interests might be harmed. Despite this concern I have published some such reports in the past (Zynga Analysis, Diablo 3 real money auction house analysis, and a Guild Wars 2 economy analysis) when I felt the public need was great enough. There has been a lot of demand from readership for just such a report that would be helpful in understanding the current F2P environment and how products perform inside that environment.
This analysis only focuses on F2P mobile tower defense games, in order to make the analysis simpler. All of the issues I discuss here can be applied to other genres and products, even if the connection may not be immediately obvious.
Here I detail the product names along with (developer name, level reached in game, dollars spent in game):
Clash of Clans (Supercell, L66, $0)
Boom Beach (Supercell, L42, $0)
DomiNations (Nexon, L109, $10)
Star Wars Commander (Lucas, Tier 6, $10)
Rival Kingdoms (Space Ape, Tier 13, $120 but only used about $40 of that)
I have been testing each game for about a year or since essentially the first day they were released (in the case of newer products). I would describe myself as both an expert user, and also a “whale” since I have no problem spending over $1000 per year on a game I enjoy.
The contents of this paper are presented with the assumption that the reader has read the previous two papers in the series, Whales Do Not Swim in the Desert and Secrets of F2P: Threat Generation. Both were published immediately prior to this paper.
Disclosure: Nexon is the only company of the four listed above that I have worked for, but that was back in 2001 and I do not believe I have a professional relationship with any of the teams involved. A sixth title was tested and not included because of a potential conflict of interest. Any other product I mention in this report is probably a product I had some involvement with at some point, since I get around.
What is a Game?
To properly understand how these products work (or don't work), it is important to go all the way back to an understanding of what a game is in order to determine if these products are meeting the needs of consumers. This is the definition I use:
Game: A contest of skill or chance between two or more participants.
This is important to note because a major condition of whether a participant is enjoying a game or not is whether they are feeling challenged, and whether there is uncertainty as to the outcome of the game. AI can take the role of one or more participants in modern times. While “fairness” is generally an important component of how enjoyable a game is for all participants, I don't put it into the most basic definition. Viewed as a whole, 99+% of F2P products in the mobile space are not games. The ability to assure a win by spending removes both the skill and chance conditions. This is not a fault of the F2P business model, but is caused by a failure to understand the role of reward systems in games and thus a casual willingness to compromise these reward systems.
While these products may not be true games, they can attempt to “pass” as games by including various game elements. In products without human opponents, this is often done by creating the illusion of randomness in the challenges even though these challenges are typically carefully engineered and not random. The focus of this paper is on multiplayer competitive games, so a careful analysis of competitive game systems is going to be attempted.
PvV, PvP, and the Gaming Social Scale (GSS)
While I have been using the term “PvV” in my writings since at least 2002, the term is not in common use so I must start by defining this term:
PvV (Player vs. Victim): Non-consensual contests between two opponents. Since the conflict is elective on the part of the initiator, the initiator will generally attempt to find a favorable contest that they will almost surely win. The initiator is generally referred to as a “griefer”. The target of the griefer generally has no way to avoid the unfair contest (without spending real money) and is not allowed to decline the contest, thus making the contest non-consensual.
PvP (Player vs. Player): Consensual contests between two opponents where the outcome is typically uncertain. Exceptions include training exercises between two friendly opponents with unequal skill/power for the purpose of training the weaker opponent. In this latter case the outcome is generally certain but it is still considered PvP.
PvV is a strong anti-social mechanic. Cooperative PvP (where two teams square off against each other) is a strong social mechanic. Individual (1 on 1) PvP is a weak social mechanic.
All five products I tested utilized PvV as their primary gameplay mechanic.
The quality of social and peer interactions in a gaming product is the primary indicator of commercial success of the product. I am of the belief that a big part of this is the hormone oxytocin, which I believe is an even more powerful reward chemical than dopamine. While I am talking with various neuroscientists interested in this subject, the research in this area still has not been done so it is reasonable to treat this statement about oxytocin as opinion, not fact.
To facilitate discussion of gaming social mechanics, I have created the Gaming Social Scale (GSS), which follows:
Gaming Social Scale
Class 1: Gameplay elements that rely on persistent PvV engagements.
Class 2: Gameplay elements that rely on anonymous PvV engagements.
Class 3: Gameplay elements that permit anonymous individual PvP engagements.
Class 4: Gameplay elements that permit anonymous cooperative PvP engagements.
Class 5: Gameplay elements that permit persistent cooperative PvP engagements.
All commercial metrics improve as you move towards Class 5.
Gaming social scale ratings of “class 1” can be abbreviated as “GSS1”. A GSS1 game would permit a griefer to repeatedly attack the same victim. While I have seen some mobile games use GSS1, I can't describe any of them as commercially successful.
As mentioned previously, all 5 TD products used GSS2 mechanics, and in truth the genre is built around this GSS class. Tower defense implies that you have to defend against something.
A game like Hearthstone would be in GSS3 because no combats occur unless you consent to them. The combats may still be of low quality if the matchmaker is poor, but that will be discussed in the next section of this paper.
Blockbuster games like League of Legends and World of Tanks are mostly in the GSS4 category, with some ability to play with friends and make it GSS5. If you don't know your teammates, it is GSS4. If you do know them, and play with them regularly, it is GSS5. Much of my work at Wargaming was to push our current and future products over the line from GSS4 to GSS5. World of Warcraft would be a game that is squarely in the GSS5 category.
Some games have a mix of play that puts them in more than one category. EVE Online has both GSS5 and GSS1 elements. The GSS1 elements are what make the game so challenging for new players, and I would credit those elements as being the primary reason that EVE has not been an even bigger success.
Rival Kingdoms is another game with multiple gameplay elements in more than one category. The primary play is GSS2 as mentioned prior. There is also a fairly robust “kingdom battles” mode available to groups of players that join “kingdoms” of up to 40 players. An assortment of your opponents (10 of their total) must be battled through to defeat their leader. The better your team does as a whole, the greater the rewards to each team member.
Thus this additional gameplay layer is mechanically a GSS3 gameplay type (since you battle enemy bases alone) but in the metagame your teammates can follow your progress and cheer you on. This smacks of GSS5 interaction. This makes RK by far the most social of the 5 games in this group, and thus the most competitive commercially. Kudos to Space Ape!
The type of interactions in a game are not necessarily the same as the quality of interactions in a game. Players are in the “sweet spot” of engagement when they feel their skills are tested by a worthy challenge, and that the results of the engagement are uncertain. GSS ratings are very important because if there is no audience for an event, that event has much less value to the individual and the community. But, no one wants to watch an unfair match, that is just sad.
Here is where a matchmaking engine comes in. An effective matchmaking engine does a good job of maintaining fairness so that all participants feel like they are being challenged and that their skills make a difference in the bout.
Clash of Clans, Boom Beach, DomiNations, and Star Wars: Commander all allow the attacker to choose who they will attack. This is not a matchmaker. Further, they allow the attacker to view the defenses of any potential victim, even seeing if some of the defenses (“traps”) have been neutralized by other attackers. This feature pushes these games into GSS1 territory by allowing a parade of griefers (“Airplane” style) to line up and beat on a defender while they are offline and not even playing.
These games do put up a “shield” on the defender if the defender takes a certain amount of damage. A substantial amount of resources can be stolen from the defender without tripping this threshold, allowing the defender to be hit hard several times in 15 minutes. Boom Beach does not even give a shield to a defender just hit, maintaining constant threat.
All of these games reward players for finding and attacking weaker opponents. They also allow you to intentionally lose battles with just a few units to lower your ranking, and thus improving your selection of victims. This is not only cruel to the victims, but it also pulls the griefers out of the ideal pleasure zone for them by removing challenge. There is a very small percent (~1%) of players that are anti-social enough that they don't care about challenge and actually prefer to grief. These players will pay for this pleasure. This can make them show up as “whales” in anti-social games, and lead to the false conclusion that all whales are anti-social in these sorts of games because social big spenders decline to spend and thus don't show up in the statistics.
Rival Kingdoms does things differently. It has an actual matchmaker that assigns opponents to each player. It does not try to make each match fair. Instead it starts a “run” with easy opponents and makes them progressively harder until the player loses and ends the run. The longer a run lasts, the greater the rewards. This is an example of what I call an asymmetrical matchmaker which, if done properly, is even more powerful than a symmetrical matchmaker. World of Tanks is a good example of a game with a symmetrical matchmaker, where the matchmaker tries very hard to make both sides equal in a battle.
The reason asymmetrical matchmakers outperform symmetrical matchmakers is that the matchmaker essentially can react to the skill level of each player, and reward them for defeating higher ranked players. This makes the first 40 or 80 hours of play in RK really high quality. Space Ape works hard to get players to spend heavily during this period. During this time each player's “honor” score goes up as they get wins. Within that 40 to 80 hour period the player hits an honor cap of 5000 and this disables the matchmaker.
If there was no cap, players would continue to float in the ratings until they hit their steady state. That might be at 6000, 10000, or even 20000. The steady state is attained when honor gains match honor losses. By putting an honor cap in, this disables the matchmaker and forces 5000 honor players to fight 20,000 honor players. Without this honor cap, the highest rated (and presumably the highest spending) players would be forced to fight each other. Which would be really challenging. Perhaps even frustrating. My best guess as to why Space Ape would voluntarily disable their own very well designed matchmaker is that they were led by the tainted statistics I mention here and in Whales Do Not Swim in the Desert into thinking that “whales” are not the sort of people that want fair play.
The problem here is that by projecting their stereotype bias onto the big spending population, they cater to the small fraction that exhibit the behavior they are anticipating, and alienate the vast majority of big spenders. The result is a huge loss in revenue, which will show up first as a large amount of churn right after players hit 5000 honor. Space Ape continues to run accounts as “defenders” long after they quit, so there is never a shortage of bases to attack.
I go into this level of detail on Rival Kingdoms not because I think it is a bad design. I focus on it because I think it is innovative for many of the right reasons, but it would seem that someone who did not understand consumer behavior intentionally broke a well designed system.
Sure, without this design choice the 50 or so players with L16 strongholds would probably get really bored fighting each other over and over. But anyone that has spent that much money to buy a L16 stronghold is going to get bored anyways with the lack of challenge. The honor cap is not going to save these players from spending and churning.
EVE Online is an older game that has no matchmaker. The result of this design choice is that combat usually takes the form of griefing (GSS1) even while the non-combat action is highly cooperative. Players are discouraged from seeking “fair” combats, even if they would be the most entertaining, because they would also result in a high risk of economic loss. So this game that is often cited as having the best virtual economy ever designed, has a game design that causes the economy to suppress the gameplay experience.
Sure EVE Online would not be the same game with a matchmaker, but this also indicates that there is room to make a “better EVE”. It can only be done with a knowledge of what worked and what did not work in EVE. Trying to copy highly social games like EVE Online or World of Warcraft without understanding why those games perform well year after year is a sure fire way to lose money.
All of the five games I reviewed for this paper are in many ways reskins of the older Facebook game Backyard Monsters. Reskinning has a suppressive effect on innovation. If you don't innovate, you can't be blamed for making a mistake. “I copied it just as you told me!” Backyard Monsters, like many games during the brief but sensational Facebook games era, used a central building as a pay/time gate.
Raising this central building was generally extremely time consuming but allowed those “without patience” to spend to bypass the wait. Supposedly whales were these impatient people. Whales Do Not Swim in the Desert is my attempt to debunk this myth. But the myth is what all these games are based on.
One of the problems with this mechanic is that opponents typically get harder as you “tier up”, in an effort on the part of developers to maintain threat by proxy. If a player could avoid threat by proxy permanently by spending, there would be no way to keep charging this player. So any reduction in threat through spending is temporary, and ultimately illusory.
So while a player may have been feeling tough at Tier 6 when all of their buildings were also L6, the moment they raise their central building to L7 (Tier 7) they will find themselves the weakest player at Tier 7. This becomes a strong disincentive to advance through the game content.
Thus players will tend to “turtle” at a tier and stop advancing. Especially since they know the “new” content will just be a reskinned version of the older content. What is the point in advancing? It will not bring relief from griefing. Once a player realizes this they will stop advancing, and stop spending.
Turtling is a symptom of a systemic design flaw carried over from the punishing F2P “fun pain” methods using in earlier Facebook games. These sorts of primitive time/pay gates are very poor performers. It is unfortunate that in the process of reskinning, many design elements like these were carried over to modern games without much thought. In the absence of a central building time/pay gate, there is no logical point where a player will be encouraged to turtle. The whole idea of a time gate is based on false data about the behavior patterns of big spending players.
You will note that industry leading games such as World of Warcraft, EVE Online, League of Legends, and World of Tanks do not use time gates. Progress may slow as you get further into the game, but there is always incentive to advance. Some turtling is seen in World of Tanks, but this is because the game is designed to be economically F2P up to about Tier 7, and then to start to cost some maintenance for repairs beyond that. Players that do not want to spend will tend to turtle some time prior to Tier 9 or will be required to play a premium tank to generate funds for repairs of non-premium tanks. This model was improved in both World of Tanks Blitz and World of Warships, but I am not at liberty to say how. It has always been the intent of Wargaming founders to maintain a top quality play experience even for non-payers, so this sort of turtling was not seen as a negative characteristic, and the matchmaker protected these players from abuse.
I don't really mention the Star Wars game in this article because to me the game is very similar to Clash of Clans, but with much higher graphics and sound quality. It also does not use Comic Sans to pull in very young users, which I appreciate. Of course it could be argued that the Star Wars franchise (now owned by Disney) does not need Comic Sans for this purpose. I am a big Star Wars fan, so I kind of enjoy the game because it feels like Star Wars even if the gameplay is not innovative.
It is my opinion that Supercell has really earned their position in the industry by understanding the importance of gender neutrality. Women make up 52% of gamers and generally have the “power of the purse” in nuclear families. That said, I think Hay Day is aimed more at a less aggressive audience (with no direct combat), and Boom Beach is aimed at a more aggressive audience (with a military theme and no shielding feature). Clash of Clans, with it's intense Comic Sans art style (I have pink and gold walls in my base) and simplistic gameplay, is aimed at a younger demographic.
I am a firm believer in “cradle to grave” branding, where you get your audience interested in your products as early as possible. I have my doubts as to the suitability of the F2P business model for children, but I am glad that platform protections (and attitudes) have improved a lot since I wrote Children's Monetization in 2013.
Two of the five games here produced situations that were grossly unfair. In the other three games I always felt like I could at least affect another player. In Boom Beach, once a player gets the grenadier unit (difficult to reach without spending), they can attack other player's bases without the defending base getting off a single shot due to their range being greater than tower range.
In DomiNations, the lack of effective matchmaking allows Industrial Era players (with aircraft) to attack pre-Industrial Era players that are not allowed to build any sort of defenses that affect aircraft. This is an auto-loss for the defender, who can only watch helplessly as their base is razed.
I never had this feeling in Clash of Clans since the arrow towers, which you get in the very beginning of the game, can at least affect flying units.
I see a tremendous amount of opportunity for growth and profit in the mobile F2P space. From what I can tell, mobile is the platform of the future (and present). What is holding it back is a general lack of innovation and game design. I have a difficult time even describing the products currently in the mobile F2P space as “games”.
They are not contests of skill or chance between two or more participants. This is not a complicated or difficult standard to match, we had no difficulties with it before we went to a F2P business model. I would encourage decision makers in our industry to refocus on the making of games, with the intent of bringing value to the consumers they serve.