2016-03 ![[moses_a_creator-god_stands_proudly_looking_forward_to_the_futur_3f5ae440-ca48-411c-9fb4-b3b21b5ae979.png]] # Notes from Dower's GDC talk https://www.youtube.com/watch?v=1ZtlyZOGqAg Jonathan Dower, Supercell, Quality is worth killing for Wednesday, March 16, 2016 2:00 PM Talk before the talk: Jonathan says there aren't really cleanly defined processes. It's up to the team. First game at supercell -- smash land. Soft launched, killed. Then his team helped clash royale get to beta. Now on a new top secret awesome game. Outline 1. Supercell is a serial killer 2. Why killing is important 3. Smash land - a real world example of murder 4. Killing process 1. Supercell is a Serial Killer We have always been serial killers. Timeline: Gunshine.net killed 2011 Pets vs orcs killed 2011 Hay Day 2012 Battle buddies killed 2012 Clash of clans 2013 Boom beach 2014 Spooky pop killed 2014 Smash land 2015 killed Clash royale 2016 Heaps of other games killed in there as well. Since 2014 [Countless ideas killed as concept] 7 killed at proto 2 killed at soft launch 1 launched globally 1 in 10 release globally since 2014. 1. Why is killing important to us? Quality. The best people make the best games. Yeah that's corporate slogany. But internally it's super important. This, as well as communication, is the most important thing. Best people --> small teams = best games (with some luck) Best people = * Proactive doers * Passion for excellence * Good person Small teams -> * Independence * Focus * Ownership (what gets me up in the morning, gets me going to work, it's my thoughts and the team's thoughts and no one telling us what to do) = quality What is quality to us? * Fun! * Gameplay * All the fun little bouncy stuff, effects * + social. * Meaningful social interactions with people * Games that people play for years and years * Truly global * Develop for everyone * Western markets, all of Asia, * Everyone! After 2 years, 1 in 10 still play Clash every day. WHAT THE FUCK 100 million people play Supercell games every day. 100 m DAU. More now with Royale. Financial goals are secondary to quality. Another way to look at it: financial goals follow great games. Retention. Looked at top grossing games. Supercell games are at top of list. Cool. Now sorted by retention. (Source: sensor tower) Sorting by retention completely messed up the top grossing list. But Supercell games stayed there at the top. Cool. But it's not all bunnies and fairy dust. * Best people don't always make the best teams. * Being small is not always focused. * Being independent has its own pressure. Lots of responsibility there. It's all up to the team and the individuals to try and not work that out. And that often means making massive decisions for the company. * Team goals come before personal goals. Less is more (or more is less) More games + more people + more management/bureaucracy = less ownership + less independence = less quality. 1. Smash land. A real world example of murder. How big is small? Supercell is 180 worldwide. 70 developers. All in Helsinki. 3-15 game teams. New team might be 3-4 people. Live games around 15 people. Clash is around 15 people. Wtf? Smash Land team size - started with 3, got up to like 10, average 4.96. How do we make games? Shows the org chart thing. Then the cell model. 5 ish people, 1 non games person who's good at trust, is a team, all that reporting to one person. How we made Smash Land. 1. Found Monster Strike before it was a success in Japan, loved that core slingshotty mechanic, wanted to see if we can build something around that that we'd like to play. Decided on this concept. 2. Focused on gameplay first. The feeling of how the character reacted, the audio, the pops and all that. 3. Got deep into the pvp experience. Lot of technology they had to work on to get the PVP and latency right. Spent lots of time on that. Risky for Supercell. 4. Company playables. Teams decide on the milestones, then deliver it to the company, all 180 people play it. Team then sends out some kind of form. 1. Now some teams are trying to do it differently, release it to a few limited people, talk directly with them, limit the noise level. 5. Set our beta targets about 3 weeks before soft launch (run them by CEO) 6. Boom pressed the button to soft launch 1. Beta 1. 10,000 DAU by day 7 (without going hard on marketing at all, wanted to see how it would naturally grow) 2. Retention and arpdau - below our target. For this type of game we felt like we needed a lot more, because it's pvp, so esp for higher levels we needed more. 3. Session count was low 4. Players more likely to stay after reaching PvP …. Cool! 2. Be bold in beta! How do we try to fix this situation because we weren't hitting our targets. 1. Entry flow much faster. Made it 14 mins to PvP. Was 45. After the fact, realized they should have been even bolder. Shouldn't put so much focus on the single player. Single player was a big button, should have been made totally secondary. 2. Added team challenge mode. 3. New short term progression for pvp 4. New resources 5. Added items, spells 3. Nothing changed :( 1. Except… This little progression bar, we didn't have it for multiplayer, but then we added it, and it gave us a boost. 2. These kind of learnings flow into other games. (the progression bar became Royale's crown chest accumulator) 4. Killed it! 1. Not a game people would play for years and years. Or decades. 2. Beta targets not met. So didn't give it a good feeling. 3. Lack of content 1. Meta game was weak and narrow 2. Hero usage/collection was flawed 4. Updates started to feel cluttered. Didn't feel like the game we were trying to make from the start… Didn't feel right. 5. But we can fix it! (Warning!) 1. We make our games, we love them, we get emotionally connected to them 2. But at some point you have to think can I actually fix this, is my time well spent here, or should I move on and make the game I should make. 1. How do we kill games? * Team sets KPI targets * Retention and arpdau * Team checks targets with Ilkka before launch * Targets are transparent to whole company * Give a presentation to small group on what we think we can do * Then it's up to the players. Release the game and hope for the best. Blah blah, but who actually makes the decision? The team. All good decisions are made in the sauna. Release the game, weren't feeling it, went to the sauna, had some beers. * Can we lock into this game for years? * It just didn't feel right. We felt like we could do better for the players. * The second you think about killing it, kill it. Chances are you already have. * Then tell the CEO. Sent him an email. * Share learnings. * Company wide post mortem. * This is when your company really grows and learns. * We put a lot of effort into it. So what happened to the dudes? Wound up working on a bunch of prototypes that we killed pretty quickly. Lot of the guys wound up helping Royale out. Jonathan went on to a new awesome game. Finally, for any of this to work, we need…. * Trust. That guy at the bottom of the tree needs to trust the teams. Otherwise people start to lose confidence, and you risk breaking apart/breaking their hearts? * No fear of failing. * Best teams. Questions. Who is the person in between ceo and teams? It might be me, the games lead. The way I see my role is I'm the guy that gets up and talks. Every Monday morning the game leads we have breakfast and we get up and talk and say what we're gonna do this week. Every Friday we do some simple slides if there's anything interesting, any learnings. Also see: "the art of killing games" by Adam teller, woot a, euro gdc 2014. Aiming for these games that last for years is your best chance. If you're not aiming that, then I don't know what are you aiming for. I think this narrows it. How do you keep morale up when you have several consecutive games that you kill? And how does someone bring up the idea that I don't think this is working out? For the mood: I think the fact that we actually make the decision ourselves makes it much better. It doesn't feel that bad. Plus it's made in the sauna, with beer and so on. Jonathan just swiveled his chair and said, so, what do you guys reckon? And it went from there. Do you do play testing with external people? Usually quite late. Smash Land took 10 months to make, and it was in the last 3 months that we went externally with a little focus group team. We feel pretty confident with our company that we get a lot of good feedback, honest feedback about how our games are. So we can survive on that for a while. Are the targets for every team in terms of retention and arpdau? Yes. But the numbers change depending on the team, the game, what they feel the game might be. The target numbers come maybe 3 weeks before we go to beta. You guys never do cross promotion between titles. Why? We do, Clash Royale was promoted within Clash Clans in the inbox. I don't really know why we don't do it more. Probably a player experience thing. How long do you iterate? It's up to the team. It's kinda like do we get this feeling that we're affecting things? With Smash Land it felt like it was getting cluttered. My one regret is we should have been bolder in our beta decisions. Instead of just adding stuff, what if we just removed the map? Overcoming analysis paralysis. Luke Muscat at GDC 2015. And Supercell says no a lot, in different ways, at different times. Anyone can come up with an idea for a game and convince other Supercell employees to join their team. The best way to do this is by developing a prototype, which usually takes a week. From then on, the game lead stands up every Friday afternoon and tells the rest of the company how the game is progressing. Many games are killed at this stage. A few months later, the team will put out a "company playable", which every employee plays and gives feedback. Many games are killed at this stage. After this, the team prepares a date for a soft launch, in the Canadian app store (to allow games to keep a slightly lower profile). After launch, the team closely watches how the game progresses. Many are killed at this stage, most recently, Spooky Pop and Smash Land. They only lasted a few months in 2013. "It's not for everybody, and it's not for the faint-hearted," Louhento says. http://www.wired.co.uk/magazine/archive/2015/12/features/supercell-clash-of-clans-hay-day # [[Journal section]] - 2023-05-12 Fri 11:57 AM - The idea of [[Good enough]], the [[Good enough paper]] etc, could be the good enough game. - Important idea from Supercell is that you have to [[Specify your quality bar]] - 2023-05-12 Fri 12:00 PM - And to nodify the other big idea from this talk: [[Minimize time spent on branches that won't get you there]]