Monetization Design: The Dark Side of Gacha

[Music] I'd like to remind you to turn your cellphone's into silent mode and by the way this is my second talk here at GDC and it is completely the same as the first one so you may have probably heard this one okay now that's let's start my name is Vladimir I am from Russia and game development studio pick Sonic and I have about seven years of pile experience and games for social networks and I like to share with you my experience of answering specific questions at big Sonic and actually I think that these questions bother every games to do no matter what games you are working on what platform or what genrih still you have to answer them and they

are as false okay we've got some revenue flow now how exactly are we going to keep it or how we are going to increase it what should we do next how to grow to make your game grow so I understand that answering these questions could be answer to these questions could be right greatly depending on your field of expertise on the market background history of the company in all that stuff and this is just my version first of all let me tell you a bit about our context where are we here we are situated in Moscow and we've been making games since 2009 and war robots is our biggest hit so far it is a mobile shooter PvP game where players pilot giant robots fighting each other with various weapons like missiles lasers cannons machine guns all that stuff and the game was

released in 2014 it is pretty popular right now by the way this slide is outdated a bit now we have more than 100 and 20 million installs on all the platforms and about one and a half million of players pilot robots in our game everyday why the game became popular in the beginning white became successful I think that at the time when it was released they were very few mobile shooters on the market and player 6lu actually wanted to play mobile shooters and they had little choice the game wasn't very good at the time there were a lot of flaws and the metrics weren't good but still we had community our community growing we had our Facebook's grow Facebook groups growing the word of mouth about the game it was praying and we had to answer

these questions I said it in the beginning that how exactly are we going to grow we are mobile and we are free to play and we are game developers right so we wanted to add new features to the game but what features exactly and for the sake of my argument here I'll talk about two models two approaches there could be different classifications or there are a lot of way ways to classify features but I like to talk about the hype model as I call it and it is characterized by a high spike of revenue in the beginning when it is released and the quick drop down afterwards usually most of the content in the game released behaves like this the second model it is what I call value growth model and it is all about sacrificing short-term gain for long-term profit it

couldn't it could be difficult to understand if the speech was successful from the beginning because you don't have a high revenue spike from the start but it will definitely bring you money utter words basically value growth model it is a bit trickier to design this kind of features because it is about designing needs that are impossible to fulfill in the game they should be continuous and it is like the very basic human needs like the need for food for sleep for sex all that stuff so hi poodle I think it is a bit easier to design and to analyze if it is successful or not after the release which one of these models is better I don't know frankly and there are advantages and disadvantages of both approaches but first of all what you should understand that a it should

correspond with your goals if if exactly you want to get returns fast in this case you probably should use the hype model as a methodology for the designing of your feature and if you are going to if you are ready to wait to get more money in this case well your growth is much better for you basically it is like a marshmallow test for offered to kids when you have just one marshmallow and you can eat it or you can wait to get the second marshmallow and in 2015 when we thought about it we actually failed that test we ate our marshmallow and we wanted to grow quickly because we needed money for user acquisition we wanted our marketing budget to grow quickly now let's try to compare the examples of features for for each kind of model and

the example of the hype model which was made in war robots is gotcha basically gotcha is like chocolate egg with a toy or here you may notice the kinder joy you have a random reward inside the egg but even if you don't like this reward you still have the chocolate and it is useful for you every every every chocolate egg you buy it is still useful to you the third principle you can use that actually real time and in fact chores can't use this principle but we can it is creating illusion of randomness instead of real randomness what I mean is imagine that you are throwing a pair of dices and you get Snake Eyes five times in a row this is how real randomness works it doesn't

have to be fair but we humans are upset with it and we perceive this as something unfair we want it to be irregular that is why if you are designing the experience of players using random mechanics if you are creating an illusion of randomness but it is a regular all the time it would be much better for you and for players in your game so we started making releasing Gaja the first one was on Halloween 2016 and it was a part of the event at the time we had just one type of chests and players played battles earned special event currency called pumpkins and opened the chests with the rewards they were different rewards inside it robots weapons currencies and the only UX feature we've added at the time was that you could open multiple chests with just one tap that's all that we did we

introduced it on the event just to have an insurance if players didn't like these then we the event would be over and the game returned to its previous state but it was successful for us and this is why we continued we made a second iteration on the new year 17 when we've added our version of the chocolate in the kinder joy which is when you are opening every chest a special progress bar feels a little and when it is filled completely player gets super reward that is useful to everyone in the game that's how it works that's how works the fact that every opening of the chest no matter what will you receive inside we'll be useful to you the third iteration was on Lunar New Year 17 and

we've added two more chests to address the needs of our game community some players actually wanted these chests to be a bit cheaper and some players wanted the rewards to be better and we've added premium chests with a higher price and cheap chests with lower price just so that you could play it as a lottery and actually this was the first time when our game community signaled that they don't like how exactly we are designing our monetization where the game is moved and this wasn't a problem for us at the time because the metrics were good there were no problems at all and we just continued the fourth iteration was on were robots birthday and we tried to avoid adding new event currencies and from now on chests were opened with hard currency with usual hard currency of the

game in this way we've created a loop when player opens the chest gets the hard currency and they can use it to open the next chest but this wasn't a successful thing for us because players feedback was that in game hard currency was too valuable for them they didn't want to spend it on opening the chest that is why we returned to the previous state but we recreated loops with our event currencies on the 4th of July now why we experimented with this feature so many times why we did it in so many iterations the answer is actually on this chart it is our revenue chart and you can see that every time we've added gadget to our events every time it was spiked in the revenue that is why we understood that we actually want to add

these mechanics to the game permanently and we need it in number 2017 it was designed to Spall's there were three special currencies key cards red green and blue and they represented time limited seasons so that when one season ends and for instance you get compensations for red key cards in different currencies or other types of rewards and now you have to to use green key cards to open chests in this way we actually prevented key cards from stacking up on players accounts this was a very successful feature for us actually I might say that this was the most successful feature for robots in terms of financial games because we had two or three times increase of our revenue at a time and we thought that it is a new plateau for robots actually we

had now we now had the fuel to continue our experiments for with monetization and we've added more features like well dopher system I'll talk about it later and the other versions of indirect sale system because basically what gotcha is in our game it is a way to sell your content indirectly to players to hide the real price of the of this unit because player doesn't know how many chests they have to open to get this unit and we've added component system it was how it was our way to sell half of the robot from that point he could offer players 5,000 components when for instance they need 10,000 to craft a robot but the market changed and there

was a big media scandal around Star Wars Battlefront 2 and I remember that some players even joked about it like imagine if Darth Vader cost not 40 bucks but 300 bucks implying our the most expensive robot in the game we looked at it but steal game metrics were good they weren't any problems on the game metrics their attention was fine the conversion rates monetization everything was good the only problem was on the iOS rating this job wasn't very big but it was significant for us and we've definitely we've won in terms of money with this feature but we've lost something and first of all players actually reached every developer really good through social media and wrote them you're bad you suck we don't like what you do with the game and people started to be

ashamed of what they're doing it's like imagine you're working on one of the best mobile shooters in the world and you don't like what you do that's that that was a problem for us and it turned into be a slower production the features made after that were made slower because people were really afraid to get another wave of hatred from our community that was the first time for us when we understood that we are doing something wrong and we need to change this thing now to the second model example which is wild offer system basically what is wild offers it is a system that generates offers two players with discounts and it can offer different things like robots head ups weapons equipment resources all the all kinds of stuff but it these offers are generated randomly with some

some algorithm ins inside of the boundaries set by the game designer on the back end for instance a game designer determines like the discount should be no less than 10% and no more than 25% for instance or you can offer this type of robot this kind of weapon but you can't offer this type of robot and that's how it basically works the design pillars of this system were as follows first of all it was iterative approach in the beginning we don't know anything about our user and we just bombard them with offers we are firing a shotgun offers to them and you can see here an imaginary dialogue between the system and the user and at first it just offers different kinds of animals to them but at some point the system understands that this particular user likes Lions and now it tries to offer them accessories for lions in this way we get

from grapeshot shooting to very aimed sniper precision offers the second principle is bargaining it is basically if you're turning down the offer the next offer you'll get will be better it will have lower price or higher higher value but if we used only these two principles the best behavior to address the system would be to turn down every offer to decline every offer till you get the maximum value for the lowest price and that is why we've added the third principle which is randomness to avoid predictability we've inserted random noise offers to make this system hardly readable for users and here you can see some absurd things white ones are they noise offers and spelled in black are the real target offers and they adjust here to these guys our

algorithm that's all sometimes actually they work sometimes their players buy them this system of course it has some drawbacks and first of all it generates a lot of data imagine you have to save all the transaction history with this system for every user and it's quite heavy on on the back end and it's even harder to analyze this ocean of data and the best way I think here is segmentation if you divide if you are dividing players into cohorts and try to find correlations between player behavior and some offers then it will work the second problem is revenue cannibalization since the basic tool of this system is offering this car if discounts are too high then you probably will cannibalize your revenue so you need to control this thing and sometimes we it is difficult for us to understand what player needs because we know we know his payment history we know

he is transaction history with the system but sometimes it's hard for us to understand their playstyle their desires in the game maybe this player likes brawler robots or this player like sniper robots right now we have the only trigger here it is a workshop system where you can craft components for robots and we know that if a particular player crafts a lot of components of this robot then probably we can offer him more components of this robot or maybe we can offer him the whole robot itself we did some optimizations this what the wild offers first of all we've increased the frequency of generating offers in the beginning it was once per day and we've increased it to twice per day offer shown and actually it worked the second thing we did was faster movement between

price tires from low price tires too high and vice versa so that now if you buy five five bucks offer for instance you maybe you will get 40 bucks offer next just so that the system tries to make leaps between tires to move quickly and finally we've increased the cap of the initial offer when we don't know anything about our user to 20 dollars in the beginning it was ten dollars because it on average it is the most comfortable price for most of the players just statistics but when we increase it to $20 offers it resulted in a higher revenue you can see below the real revenue chart of the system after the optimization we did it was just in one release now let's try to compare these two systems in terms of revenue and here you can see the bang user

percentage the yellow chart is gotcha and our green chart is wild over system and you can see that they resemble very much the initial charts I've showed you in the beginning by the way the drop down of the weld offers afterwards doesn't relate to its design it we screwed it up but in a very different way it's completely different story now let's see at the revenue let's look at the revenue here you see Gacha system in light green and wild offer system in yellow probably it would be easier for you if I show you just to these two sources here you can see you can see that in the beginning gadget generated a lot of revenue most of the webbing of the game but later it declined because it it has lost its value to our players

what are the results of this new approach first of all our morale started to grow and not just because of making these two features we did a lot of UX improvements requested by all players in 2018 and right now our user rating is higher than ever before and that's that it was good our revenue is much more stable and predictable now since we don't have a lot of fluctuations made by the gun before and it is much easier for our marketing team and analytics to plan user acquisition for for one year for instance what we've learned from these two examples first of all long-term planning for us works much better because it helps us to build trust between our studio and investors they

see much more stable behavior of the game and actually if you think long-term it is easier to get the desired result some some time afterwards you shouldn't definitely underestimate the influence of your community because even if it isn't shown in any metrics still it will be significant after all we are all humans and for us uncontrollable growth was not a very good thing we wanted to grow of course and it helped us but it's hard to plan long term budgets and it's hard to plan long term development of the studio if you're revansch art is very fluctuating like this well basically what I wanted to say here is not just gradually though

no you can use gosh I know a lot of games where grass is designed in a way in a way according to principles of really growth actually I just wanted to demonstrate these two approaches and to tell you that you always have to understand the consequences of which approach for instance we could do the same we could design gotcha allowing along with belly growth principles but in this case probably we didn't have as high revenue spike as we got because in this case we couldn't squeeze so many payments from our players so that is all thank you if you have any questions please ask oh there is a mic yeah I had a question so your wild offers uh-huh so what we've seen is like in a lot of games where you

discount your currency and your offer is that like over time people kind of just devaluate them and you have to constantly be creating more yeah so it is a problem with your wild offers how do you go about kind of managing that like what are your thoughts about how you how you how much you discount and how often you discounted okay what I found is that in game currencies lose value to players if you do it regularly and player knows for instance that okay I didn't have to buy it right now because I know there will be a sale next month and I'll buy it them if it is more spacial more random and they don't know what exactly will happen next months in this case it is okay it doesn't devaluate I think sure so I was

wondering this is really fascinating and you mentioned one of the big problem for gotcha is the backlash coming from the community but you know sometimes people backlash when they realize they are paying different prices you're the same item in the game as well did you see anything over there because I think it's it's a really clever get around of the issue of hiding the price in your steel you're essentially having price discrimination making sure people are paying what they're going to pay but do you see people responding negatively to that or they're generally not really talking to each other about oh you got $5.00 why do I need to pay 10 do you mean that people get different offers and that is why they could be upset so that they don't have enough for a particular for instance their friend yes this is the problem and it it relates a lot for some kind of you know payment tourism for if you differentiate the prices between

countries then a lot of players will they have a way to deceive the system and to change the country to get lower prices but if it is not related to anything like the country or the language or something like that you can always say and that's what our support guys say that this is the complex algorithm it takes some time to to get you the right offer just wait you'll get it and mostly it works you just are not familiar with war robots that's like you went through this slide that I thought was really interesting but you went through it pretty fast where it's like before like basically you were so like gotcha for hard currency for a period and then in the next iteration you you I forgot you with three different currencies can you elaborate on that and just like how it works exactly yeah key cars yeah the key cards like so

how did the key cards like what they were they are still earned from battles but they change one after another red key cards are compensated and then you have to earn green key cards because actually we didn't want our players to stack up a lot of key cards on their accounts because it would be difficult for us to sell things in the future that was sort of similar to my question was I was curious to know how you designed that kind of the token and economy of these different soft currency xor's is the hard currency so can you explain a little bit about the thinking and the design of how many of these currencies you had I mean you'd spoke just right right now about how you didn't want people to kind of hoard these things have you built sir mechanisms where people have sort of incentivized to get rid of them and then you mentioned also

that if players understand that these currencies kind of come and go that they lose value over time so aha so so can you that that's curious to me because the game we're building is very heavily influenced on currencies and and things like that so we'd love to hear your thoughts and your design around the soft currencies and sort of had that fit into the hard currency sided thing and in your revenue model aha you see the only purpose of key cards was just opening chests so it doesn't lose value because you can't spend them on anything else I see so they're not tradable and you could they were acquired through grinding just playing the game yeah you can acquire them through grinding but if you don't use them for we have a season like a couple of month I think I don't remember ok but if you're using them if you don't use them when the season ends

you get the compensations in other currencies for that key cards so even if player doesn't use this they are still available to them ok I'll look to ask for questions later uh-huh Thanks hello have you experimented with so ends of the offer that is available to user after he receives it what is the optimal time and same question goes for the amount of offers that is available to the user at once yeah we we've experimented with the length of showing the offer at first it was just one hour because we wanted to make it some to create some time pressure but actually it didn't worked well and right now it just has offers shown for half of the day and then they are changed with next offers so for us temperature didn't worked well especially because they were

the the timer wasn't related to player opening this screen of offers but it was related for player to opening the game and in this way actually some offers players didn't see them at all that was a bug we fixed it but so right now we just we are showing this offers for half a day and then change with another yeah hey so great talk I was wondering though you mentioned a couple of what seemed like really high price points yeah $40 or $300 yeah I was wondering did it seem does it seem like you make a significant portion of revenue from those handful of players who can spend that much money or is it mostly coming from players who can spend a lot less we aren't unique here on mobile market and most of our money comes from people that

spend a lot they are frequently called whales but that's not a very good name for them I think they these people treat your game as a hobby and we usually if we like something we are ready to spend a lot of money on it yeah so I think we have most of the revenue from people who spend more than $1,000 in our game Thanks okay that's all right okay thank you for coming [Applause]