We launched on the 14th of October what we call eRepublik V1 (our public version) as opposed to eRepublik beta (that was invite only) on the 14th of October.
This was a very hard pre-launch and so was our public launch on the 21st of October. Things are just starting to get back to a manageable level (ie: the 21 people in the team get to sleep 6 hours instead of 3 hours per night), so I can write a few lines about it to share some of the stuff that happened and some learnings.
The first learning is that I agree with my co-founder George, we launched V1 to soon but it was the right thing to do.
I won´t go into the reason of why it was the right thing to do and all the other very good things that we did because I’m of the school that you learn more from your mistakes so I guess its more useful to share those.
eRepublik beta had been going just under one year (since we launched a simpler version of it on November 20th 2007) and although we had kept it “invite only” to manage its growth, it quickly had become much bigger than we expected. eRepublik beta wasn’t really built to scale, since it was a simple prototype to get a proof of concept, gather feedback from the community for what the fist step of the real product V1 should be like and start building the new world.
Just to throw a few stats, we had 35,000 citizens that visited the site daily with their inventory of virtual goods and a history of achievement and actions going back months, over 3500 companies, 13,000 newspapers and over 150,000 pages of content created by the community in the beta.
All of that had to be transferred to a new platform seamlessly. Needless to say we had more than a few hick ups in the transfer with many users loosing part of their inventory temporarily along the way. We also had many bugs we had not spotted or thought we had fixed in the testing phase that appeared when we opened up V1 to the 35,000 citizens. This, along with server fine tuning issues that meant that the site performed poorly for a few days made for a few hairy moments, where again most of the team was to use my favourite expression, “running around like headless chickens” just to fix stuff.
Now for those of you of have been involved in Internet start ups for a while, you know all of this is pretty normal and that although it’s possible to minimize the above issues it is virtually impossible to completely avoid them.
The thing is I believe that no matter how much planning you do, the shit is always going to hit the fan and then it’s a question of how quickly and efficiently you can clean it up.
But it’s a little more complicated when you are dealing with a community, as we are with eRepublik and there are clearly a few things that we underestimated or could have done better.
1) We should have communicated more: We really value our community, we know how important all the citizens are. We are building this new world for them and with them. We are providing the vision, concept, basic tools and functionality of eRepublik, but they are the ones creating the content and showing us what stuff they want and in many cases how to improve the game. In certain cases with external sites or our upcoming API, eRepublik citizens are even building themselves improvements to the game. So what we did is, 2 months before the launch we started telling them what was going to be in V1 so they could prepare. One week before the pre-launch we let 100 citizens from the community test V1 for final de-bugging and extra feedback. A lot of the 100 people in the test phase wanted us to change major things there and then and delay the pre-launch.
a. With our testers: Our first mistake was that we did not explain to testers well enough why it wasn´t possible to delay the pre-launch. We were to busy fixing and correcting things based on their feedback. It wasn’t possible to delay it, because we actually needed to get V1 out so the team could concentrate 100% on improving it, rather than manage 2 sites at the same time (V1 and Beta). Plus we had other external pressures I can´t talk about here.
b. With enough resources: When we did pre-launch and the problems 100 people can find became the problems that 35,000 people can find, we went into fixing overdrive. This was the right thing to do and I want to thank again the team for all the sleepless nights. The community saw how quickly we were fixing things and the direct feedback we got was priceless, a normal game would have probably needed months of testing to gather the information we gathered in days. But we waited to much time before acknowledging the community feedback in a coherent manner and letting them know exactly what we were doing and fixing. As a result many citizens made it very clear they were unhappy with that. When you do such a major shift one should expect that there will be a crisis and a crisis communication plan with the appropriate resources is therefore in order. We are now communicating with the community on a daily basis about the transition and things are much better, still we should have started earlier and planned this better.
2) We underestimated how attached the community was to the eRepublik beta version: V1 had to be a big step from beta technically and functionality wise. It didn’t necessarily have to be from a design and UI perspective. In order to prepare for the mobile version “anywhere any device” and accommodate some of the necessary new feature we did need to make a few major changes and in the process broke a few rules we shouldn’t have.
a. Fact there is nothing better than live fine tuning doesn’t mean you can do away with focus groups: Our main design objective was to make V1 easier to understand and simpler in appearance (although in reality there is a lot more stuff hidden under the hood). I know it will be much better when we fine tune it and we did the right thing medium to long term. We had 10 months of fine tuning with beta so clearly our first throw of V1 wasn’t going to be as good from day 1. Still more use of focus groups would have helped iron out some UI and design issues that we should not have had in pre-launch, although that is always a trade off with time.
b. Change as little as possible in appearance: Also specialists keep telling you that you should always try to minimize visual changes and UI to something people already like and use. I always had that advice at the back of my head but clearly should have made that point clearer with the team. Also we were struggling with how to make the experience familiar to old users as well as easy for new ones (something beta was clearly not). With insight we went too far in the design changes and will now have to work based on our community feedback and the help of focus groups on the design not just for UI but also for atmosphere. There a few no brainers of course that we already implemented and others that we are still working on.
c. Only do the “revolution” if its your only option and if it is (it was in our case) try not to break what ain´t broken: Good news is that the community is getting familiar with the new design, we are improving it quickly and from now on its all about natural evolution rather than revolution in that respect. There were also some modules that we removed (mayors being the biggest one) because we felt we had not designed them well in beta and they were not used much, but then realised that people that did use them cared deeply about it. We felt the right solution there is not to just give them the same thing again but to give something that is really well designed and useful so it works for them but also for many more. Still in the first instance of V1 we could have just left it there until we had something better. We broke something that did not need to be broken.
3) Communities don’t like surprises: 90% of what came in V1 we shared with the community and was welcomed because it is things that the community had requested. The other 10% we kept secret. We didn’t keep it under wraps to protect any trade secrets (although some of it is quite innovative), the main reason we did that is we felt we had found the solution to one of the main downsides of the beta and wanted to surprise everyone with it.
a. The Trivia: The main secret we kept was the fact that we gave the option of replacing all “repetitive tasks” such as clicking a button for working with a trivia (that you can skip) that consist of five questions that are contextual to what you are doing in the game and give a skill based bonus to the default result of your action depending on how many questions you answer correctly and how quickly you do it. Now this does replace a repetitive tasks with something that can be fun (as long as the questions are good and contextual, not always the case right now), requires skill and can also help you learn something new. So that is good. Also if you look at the market share of online games, a whopping 50% is online puzzles and Trivia so this seemed a good way to make eRepublik interesting to a wider audience. Finally, 90% of eRepublik users are male, it’s a strategy game after all but we are keen and making it more friendly to women and we know that this goes through more social & personalisation options and that Trivia is extremely popular for women. What happened is that a big majority of new users love the concept of the trivia, although they feel and I agree with them that it needs work and we need to improve considerably the quality of the questions, something we have already started to do. On the other hand some of our most influential citizens just hated it and basically told us to take it off or they were walking out. Now as most people who manage communities know although a lot of “hard core” users can often use this as a tactic to pressure you in going the direction they want but not really mean it and not having really the support of the majority of users, this is scary. In the end we simply couldn’t scrap the trivia, it’s just to ingrained in the gameplay (plus the first step of something larger) and although very few users actually walked out (our user base growth actually accelerated) I’m sorry for the ones who did. We should not have surprised them with this because their problem was not really just with the trivia concept, they were mad because we did not tell them about it in time for them to give us the feedback of what was wrong with it so we could make it better (ie: for example implementing it in the new war module in a way that did not reduce player vs player confrontations). We are now getting that feedback and improving it progressively plus looking at were it should really be. But had we avoided the surprise, we would probably not have lost some important users and launched a better product.
4) Don´t call it V1: MMOG’s such as eRepublik have numerous advantages over classical games that you buy in a box or download. One of the main advantages is that they really are internet web sites and as such they can be updated, improved on, fine tuned almost on a daily basis as long as your code is clean and you have good processes. This means that your product is always a work in progress and not a finished product. Of course there comes a moment when you can’t call it beta anymore and your platform must be free of large bugs but that is clearly not right after a major new platform launch. There is a lot more stuff that we learned but this is the stuff I think might be most useful to others and I hope sharing this will help you avoid repeating some of the mistakes we did.
In any case I wanted to give another big thanks to the team and our citizens because in spite of all this we still managed to fix most of the issues and we now have close to 60,000 active citizens in eRepublik almost double the number we had 1 month ago as well as a very clear vision of what we need to do going forward.
Alexis Bonte 