Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Some requests and discussion on the state of Unity analytics

Discussion in 'Unity Analytics' started by aaronflippo, Feb 8, 2019.

  1. aaronflippo

    aaronflippo

    Joined:
    Jun 7, 2012
    Posts:
    45
    Hey all, we have been using Unity Analytics for a few years, and now that we are making a mobile F2P game, we're trying to use it for in-depth analysis.

    We keep running into limitations on our ability to slice data and analyze it, and I'm wondering if any of this is on anyone's radar. It feels like Unity analytics is severely hampering our ability to iterate quickly and accurately measure our changes. Most of the serious F2P developers I've talked to are using other systems or building their own, and at this point I'm seriously considering whether we should ditch UA altogether and move towards a more robust solution.

    Here are a few issues that I feel are holding Analytics back from really being useful for us

    1. Segments are cumbersome to setup, and only apply to new data

    Sometimes we don't know what data we're looking for until after we've done some playtesting. Let's say I want to ask a question like: "How much did our 1Day conversion change among players who started on our new version 1.02?" To answer this, I need to:
    - Register a new event, play a bit so it shows up on the backend
    - setup a segment using these event parameters, and wait 24 hours for it to actually start working
    - Put out a new build, wait for a few days (or sometimes weeks) to gather enough new data to make good comparisons.
    - Squint and look at graphs showing daily trends, since there's no way to actually just get averages or totals for a segment.

    Solution: I feel that if we have events registered, we should be able to retro-actively setup segments and analyze that data.

    2. Segments can't be combined when examining data.
    Let's say I want to look at something like a level-progression funnel, but filter it to the following: "Users on Android who are also in the US". Or as another example: "Players who have never entered the store screen who have been playing for more than 3 days." You simply cannot get at this data right now.

    Solution: I feel that anywhere you can specify a custom segment currently, you should be able to query based on robust combinations of segments.

    3. Revenue doesn't filter out test purchases
    Actually, this is a very general issue: There's not a good easy way to "blacklist" development devices from UA, so it's possible to add all kinds of noise to the UA data across the board.

    Solution: We'd like to be able to prevent data from developer devices from showing up on the backend, or to remove all data from certain device/user IDs retroactively. Not every game has hundreds of thousands of users, particularly in early closed betas.

    4. A/B testing is promising, but feels like it's in early alpha mode.
    This feature sounded really promising to us when it was announced last May. However, The only thing you can measure is Day1 and Day 7 retention. This is useful, but let's say I want to measure IAP conversions, or a particular funnel conversion?

    Solution: We should be able to choose any metric UA can measure as the goal for the experiment, including event custom parameters.

    5. It's impossible to answer very basic questions about averages and totals over a time period.
    "What's the ARPU among players for December?"
    "Did session length go up with version 1.02 or not?"
    Questions like this seem very basic, and it's amazing to me that we can't answer them right now. Graphs are great, but we need to be able to aggregate.

    Thanks for listening.
     
    Last edited: Feb 8, 2019