When managing a business, we should ask whether to use a third-party tool for tracking our key performance indicators, or instead build our own system. In this article, I share our experience tracking data across various platforms and countries. We found that conventional analytics tools could not manage all our conversion funnels at scale. So we built our own solution. I outline below the challenges we faced and how we solved them.
All companies calculate, or should calculate, their business metrics in order to optimize the conversion funnel. When we launch a product in a specific market, we concentrate on analyzing cohorts for that market. For example, we analyze by week to measure whether changes have a positive impact.
In our case, measuring multiplayer game performance in real time proved very challenging. We faced numerous challenges, which I explain below.
Problems and Challenges We Faced
Multi-dimensional data: Every piece of information requires a database query. If you analyze 20 to 40 countries, 12 KPIs per country, and 9 products per platform, you end up with more than 3,000 data points to extract and analyze. Nobody can do this manually.
Time is crucial: You need to analyze critical information efficiently when making decisions. The longer you take, the longer your funnel stays unoptimized. Therefore, performing quick analyses helps improve conversion ratios.
System saturation: With all information in a single database, the system eventually saturates. In other words, query response times increase to the point where they affect the gaming experience. This forced us to migrate our analytics to a separate database.
No ROI visibility: We knew what users did inside the game, but we lacked information about acquisition costs. Additionally, users came from different sources (Google Ads, Facebook Ads, affiliates, SEO, virality). Calculating ROI per acquisition source required very complex cross-analysis.
How We Solved These Challenges
To address these issues, we built a custom analytics platform. We made the following key decisions:
Separate database: We moved all analytics to an independent database. This prevented performance degradation for our players.
Pre-calculated KPIs: Instead of running calculations on the fly, we pre-calculate everything daily. Every morning, we have all relevant data ready to review.
Custom dashboard: We developed our own dashboard that combines game data with acquisition data. This gives us a single, unified view.
Acquisition source tracking: We integrated all acquisition channels into our system. Now we can calculate ROI per source and campaign accurately.
We learned many valuable lessons throughout this process. In conclusion, building a custom analytics platform demands significant effort. However, when your product reaches a certain level of complexity and scale, you absolutely need it. I hope our experience helps you make better decisions about your own data infrastructure.