During this year, our tech team has been working hard on updating our tool which will bring some major changes to GameAnalytics. We’ll soon be launching a replacement of our backend which will introduce many new capabilities, such as multidimensional filtering.
This change is coming soon to our product (2018 Q3).
To keep things simple, we’ll describe our current backend as legacy, and the replacement as our new backend. We’ll share with you what you can do with multidimensional filtering going forward, and how this compares with our legacy backend.
For more information about these changes, take a look at our previous posts:
How does filtering work?
As you know, filtering data means choosing which parts of it you want to look at. In our tool, this normally works by picking a dimension type (like Country) and then fine tuning the specific values on what you want to see (like only viewing US & UK).
Our legacy backend is pre-aggregating all possible metrics & filters each day. In order for this to work smoothly, we put in place limits on the amount of possible combinations. This means that:
- Only one dimension type can be applied as filter (for example, country or platform)
- Selecting several values for a dimension type (like 4 specific countries) will split the data by each value automatically. Therefore it is not possible to get the filtered data in one line-chart.
Let’s break this down:
Head into the GameAnalytics tool (soon to be legacy), and select dimension type = OS Version, and then these specific values = iOS 10, iOS 11
Our legacy backend will not allow another dimension type to be added (like Country and OS Version). Also the filtering will always split the data, so a widget will display 2 individual line charts.
Here’s an example of how it would look with a country breakdown.
These limitations were the primary reason we decided to introduce new technology, allowing more flexible filtering.
What is multidimensional filtering?
With multidimensional filtering, you’ll be able to add as many dimension types as you like (like Country and Platform, or Device and Build…), and then be able to add as many values that are available for these dimensions (like selecting US, UK, and DE).
This will result in 1 line chart being displayed containing all of the data matched by all of the filters applied. It will not automatically split the data.
How to use it
Let’s say you wanted to create a filter that includes certain iOS devices and players who live in the US. You can set this up by:
- Select dimension type = Device, and then press all of the iOS devices needed
- Go back, and select dimension type = Country
- Press United States and hit apply
You can see the multidimensional filter we just created in the below image, and the data is displayed all in one line. This will work on all dashboards, as well as the explorer tool.
Split your data
If you want to see the values selected in your multidimensional filters as their own lines in your chart (like our legacy backend), then you’ll have to use our new split capability.
This is possible by clicking the split dropdown and selecting a dimension type (like OS Version).
Let’s see this in an example. Say you want to view your US audience and then display a line (split) for each specific iOS version. To do this, you’ll have to:
- Add a filter with dimension type = Country, with the value = United States
- Add a filter with dimension type = OS Version with the values = iOS 10 and iOS 11
- Lastly click on the split dropdown and select OS Version.
You should now see 2 line-charts in the widget.
As mentioned, the split capability allows you to split data by any dimension type and it’s values. What’s great about the advanced split however, is that you’ll have complete freedom to define filters for each split (line-chart) within your graph.
To do this, you’ll have to:
- Select advanced split, and then press plot. You should now have 2 plots available.
- Add filters to your first plot point.
- Add filters to your second plot point.
You’ll notice that as you update your plots, your graphs will automatically update too. Remember, this capability will work in all of your dashboards, as well as the explorer tool.
To give you a better idea of how this works, take a look at the above image. You can see that by using this feature, we can define 2 advanced splits with (iOS 10 & UK) and (iOS 11 & DK). This combination can only be done using an advanced split.
This new feature will give you the freedom to build and compare user groups, so that you can achieve insight needed for your particular gameplay.
How is this possible?
This new flexibility is possible due to our new backend being a data store instead of serving pre-aggregated values. Read our previous blog post to learn more about this new technology.
Over the past 4 years we’ve been serving you data through our legacy backend, and as it’s been pre-aggregating it’s allowed our legacy system to run quickly. Our new flexible backend will perform an actual query on a data-store, and therefore it may take a little longer to display metrics (depending on time range, size of game and complexity).
We’ve done a bunch of testing on this, and found the difference to be minor. As this is a large system change, you may notice a couple of initial rough edges, although we’ll be keeping a close eye on the performance and we’re always working on optimizing the system.
Can’t wait? Try the filtering now!
We’ll soon end our beta, which means the new backend features will be the default in our tool. Before that happens however, you can still try out the multidimensional filtering by navigating to the beta features on the Explore and Dashboard feature.
While you’re trying out them out, we also recommend that you install our latest SDK versions which have new tools available like Experiments and Command center. Make sure to check out our other product updates related to this migration, so that you’ll be up-to-date on what’s happening:
Not a current user? Don’t sweat it, we have a free demo game in our tool for you to play around with: Log in to GameAnalytics here.