Move fast, ship things: shipping 300 game titles in 90 days
OVERVIEW OF AGNITUS
Agnitus is an award winning educational startup / app for preschoolers with over 3M users and a place into top 5 on the apple store. The agnitus app built on common core standards is a complete educational system with 100+ educational games,books, assessment and planning dashboard for teachers/parents and feedback report cards. Funded by social endeavors of Eric schmitt & social capital of Chamath Palihapitiya
- Why ship so many titles? Increase main agnitus app’s traffic by creating an apps network which funneled users back into agnitus app
- Create a network of 300 games on iTunes & android each in 90 days time.
- Lead of Game Production
- Game Producer / Product Manager
- Product Researcher / Analyst
- Research: Identification of games that could be replicated easily. Identification of popular themes that work.
- Creation of an advertisement wrapper: An advertisement infrastructure that can push different ads on different games and measure them.
- Design team process: Creation of a process for a design team, which pushes towards optimum productivity, by intelligent asset creation.
- Development team process: A development process that leverages each developer’s strengths and ensures a smooth transition of work from design to development to deployment.
- Production & Deployment process: Devising a production & deployment process to ensure a smooth transition between departments, deployments on time without delays and with the right ASO.
- We were able to push 600 titles combined on the stores smoothly on time.
- We even had assets for 50 more reskins in the developers bin.
- We were able to get 500–1000 clicks daily. But not for agnitus main app. Clicks worked more for other advertisements and satellite apps.
- Were unable to convert satellite app users to agnitus customers. Hence, we were unable to create the growth system that we wanted.
RESEARCH & SETUP
Identification of games
We didn’t want to release all our games individually, we kept some just for the main agnitus app. Some games couldn’t be released individually because of dependencies on the main app. Finally, we didn’t just want to release 50 odd activities that made the cut but release multiple themes of the same game. So 30 games were finalized which will go in satellite apps production.
By doing a data analysis of the app store, we identified themes more popular in the children’s category. We narrowed down the themes to categories: Events & holidays, professions, gender based, scenes and scenarios e.g Christmas, Halloween, vehicles, dolls, jungle, under water, the city, space etc.
Game design changes
The agnitus games were intelligently designed to work together, so the leveling systems were dependent on different games. The leveling system had to be changed (for every game) to work individually. Some level progressions were changed simply with a hack in the code others had been designed again.
Creation of an advertisement wrapper
We didn’t want to show adverts to children while they were in the game. But to parents, so we put the agnitus ads at the start of each game session and when the parents wanted to access the dashboard to learn about the children’s playing behavior. A brilliant ad serving mechanism was created by the app side dev team. So we could pick and choose to add delivery right from the web to each app and have different ads in each app for A/B testing to see what works.
In additions users could simply pay a premium to get rid of the ads.
Game updates for easy theme changes
We needed to change the assets of a game as easily as possible to make sure we shipped the games on time. The game team worked on each game to separate the assets from the code, removing any dependencies or any name specifications.
It took two weeks to design the process and do our pre production. We had 65 total working days, 10 went in planning, we had 55 days to ship 300 titles. Which meant we had to ship 38 games each week to reach our goal.
We created a team of 7 designers, 10 developers, 2 animators and a team of 7 game producers.
Production & Deployment process
Production worked on weekly sprints. On Mondays we picked the games ready to go into dev phase. Each of the 7 producers was responsible to ship 5–7 games per week. Producers were judged on shipping, quality of the game and numbers of downloads of the game. Shipping was the only official criteria. But by having numbers of downloads, it pushed the producers to work with designers and developers to make better games which the audience liked.
The production process worked like this:
1- Pick games from the designer completed list in a round robin fashion
2- ASO for the marketplaces
3- Put names, descriptions in the system
4- Create package info for the developers
5- Take and create screenshots for games
6- Record video of the game for marketplaces
7- Work with developers and test builds
8- Ship all games on Friday before day’s end
So each designer had to produce assets of 42 games in 55 days. But wait we can’t have designers work till the end. Games will be going in production the first week and the last so that’s 10 days gone. So we’re down to 42 games each in 45 days. This was only possible with some clever work. So we did two things:
- Identify games where we can reuse assets. If there are three games on numbers reuse what you can. If you have made a theme and with a tweak it works on two games, that’s great.
- Have designers compete. There was no direct competition and there were no rewards defined. But we made a massive excel sheet where each designer added the games he’s completed.
After few days some designers were way ahead and slowly everyone started pushing themselves finding ways to increase their speed and make assets more cleverly. At the end all the designers finished almost at the same number with one or two exceptions. Just to have an idea how great it was at the start there was a difference of up to 400% in the output of the fastest and the slowest asset producers. And without any outside incentive the competition worked. We were done with designs within 35 working days.
The process worked like this
1- A designer will pick a theme from the theme list sheet.
2- The designer will pick the game he/she wants to work with.
3- You cannot block a theme or pick more than three themes in advance.
What the last point did was make you not only push yourself out of the comfort zone but also work faster so you finish a game and pick a theme you like for a game before it’s gone.
Unlike asset design it wasn’t feasible for every developer to work on every game. Because games were written by different developers and it’s easier to work on your own code. The process worked something like this:
1- Identify games with games lead that are assigned to you.
2- Work with producer to switch to new theme.
3- Work with animator to update animations.
4- Create builds give to producers.
5- Mark your game.
Like the design process, a single all can view sheet was created for developers to create competition among them.
Streamlining & automation
Meanwhile the app team worked on automating the production process, the design sheets were changed into online forms where designers picked the items. The producers found the items in their inbox and assigned them to developers working with them.
All the ASO items, descriptions, screenshots, builds, keys were directly put into the system online. So nothing was missed.
And the ads were also served directly from the system. This was the key to getting it all out without messing too much.
LEARNING & INSIGHTS
- Educational app users and games users are different. You can have the game user to remove ads and pay 2 bucks for it. But it’s hard to get him/her on a larger deeper application.
- Converting users for the same application is much easier than pushing users from one app to another and then converting them. The conversion rate goes too low.
- Competition within individuals among individuals for a short burst can increase productivity provided quality standards are predefined.
- Having tight shipping deadlines pushes designers to focus on improving their design pipeline, even the one’s who generally get lost in the details.
- Nothing should be oral, everything part of step of the process needs to be logged, checklists to be maintained, checked and double checked. Once the team develops the habit of following the process and trusting it things go faster not incrementally but in multiples. As long as technology is aiding not hindering.
- If you can get everyone to focus on their part and only there’s and remove all ambiguity from the process you have happy teams. Pre plan everything, don’t manage but aid the process.