How to build your analytics stack and not get a full head of grey hair
Disclaimer: this article is mostly for newly-established game companies and start-ups that are building their analytics systems. “Oldies”, read on but don’t get salty please 😀 . Also, there is no magic recipe for an analytics system. The way you end up building it depends on your circumstances. Everything below are just guidelines and personal opinion.
Quick question – is your game in development phase? If yes, good job on starting to think about analytics in advance. If not, shame on you, but keep reading.
So you are happy with your game, it’s got potential and you are thinking about some fierce analytics on it because data is cool and helps everyone. Your knee jerk reaction might be: “HIRE A BUNCH OF ANALYSTS/ DATA SCIENTISTS, QUICK, BE DATA-DRIVEN AND BUILD EVERYTHING IN HOUSE!!!!!” Ok, breathe. In the beginning, you can do a lot with Firebase analytics, Excel and only one person, and here is why.
Reason 1: Money. Hiring a lot of people & building your own analytics system is costly. If you are a new company, you are unlikely to know how much you will profit from this. If you’re really ambitious it might be worth investing in your own analytics system in the future but hold your horses and follow me to Reason 2.
Reason 2: In the beginning, you have only basic needs. If you are not in soft launch yet, your task is to spec analytics events, make sure attribution and other necessary SDKs are implemented and function properly. Also, you need to be there during playtests to check that everything works. You do not want to start building your own analytics system and nailing down all the definitions at this point, trust me. How hard can it be? Weeeeell….Try to nail these definitions down fast and I will give you a standing ovation:
- Installs: Do you know if you consider ‘first app open’ an install? Does a user count as a new player if he/she just made it past the loading screen? How do you deal with people using multiple devices?
- DAU: What is considered an “active” user? Does he/she need to only open the app or fire some events to be counted as “active”? Which timezone are you using? It can make a difference!
- Revenue: Oh this can be a topic of its own. Are you reporting verified revenue only? Do the numbers match between the store stats and your own reporting that you have just happily built? Are you sure you are taking VAT and currency conversion rates into account? How do you plan to track refunds?
- Retention: Will you go with 24h or calendar day base for your retention metrics?
- User ID: What will you use as a user ID – device ID, vendor ID, some hashed ID of your own?
And, many more. Do you want to start stressing about this while being in development phase/soft launch? Probably not.
Reason 3: Time, which is similar to Reason 1. Whether you are in soft launch or not, you’ll need to make fast decisions, and the last thing you want to do at the time is fixing data mismatches and restoring a fallen analytics system. And, if you start building your own analytics platform too early – trust me, that IS what you will be doing.
I hope I convinced you. In practice, you are likely to need different analytics solutions at different phases of your company’s life. Follow the stages below and you will probably reduce your stress levels significantly.
Stage 1: Do not bite off more than you can chew. Use a free version of Firebase analytics (or Flame plan, it is not expensive) at first. It has a lot of nice built-in features + an A/B testing solution which you WILL need. Once you have launched the game and done some initial tweaking, decided not to kill it, started actually making money – follow me to Stage 2.
Stage 2: Access your analytics potential given the revenue stream you are generating. How much will you benefit from investing into another third party analytics solution vs the cost of this investment? Will hooking up a third party tool with more functionalities be beneficial to your company? Make a good old business case and evaluate the outcome. Your analyst should be able to do it. I urge you not to think about building your own system at this stage. Maybe you are fine with Firebase? If you feel like you need more, read on.
Stage 3: Go window-shopping for solutions. The business case that was made earlier should give you an idea about the price you are willing to pay for a third-party tool. See what is on offer – Amplitude, Periscope, Looker etc. Also, evaluate the ease of implementation and set-up. Ideally, it is awesome if you can strike a deal with a promising analytics start-up or an analytics company that wants to branch into gaming – you are likely to save some money and may have an influence on the features that will be added to this third party’s analytics system in the future.
Stage 4: Make the best of your third-party tool!
Stage 5: Decision & evaluation time! Using a third-party analytics provider will help you understand your needs without having to go through the process of building the stack yourself. Maybe you feel like you require something else, e.g. have more freedom on defining concepts, re-running and backfilling the data etc. Maybe you are OK with your third-party tool – it is perfectly normal and there is no shame in it, no matter what other game analysts tell you! What matters is what gets the job done for you! How to build your own analytics system without going grey prematurely is a topic for another article ?.
