Hi Creators,
Today, we’re making two analytics updates to provide more transparency on how your experience is surfaced by recommendations:
- Qualified play through rate to track impressions to qualified plays, which measures deeper engagement than plays.
- Similar experience benchmarks to help you compare your KPIs with similar experiences that your users also play.
Qualified play through rate
Last year, we added a home recommendations conversion rate chart that measures the percent of users who play your experience after viewing it in recommendations.
Since then, we’ve updated our recommendations system to better surface experiences that drive deeper engagement and reduce clickbait.
Specifically, in recommendations for home and search, we’re now using a signal called qualified play through rate that tracks the percent of users who have a qualified play in your experience after viewing it in recommendations. A “qualified play” refers to a user’s intentional engagement with your experience, as opposed to accidental clicks or quick bounces.
With this in mind, we’ve replaced the recommendations conversion rate chart with a qualified play through rate for recommendations chart in Acquisition analytics. We’ve also added benchmarks for this metric based on similar experiences that your users also play. If your qualified play through rate is below benchmark, consider using our new funnel events to optimize your new user onboarding and testing new thumbnails that match your latest content updates.
Benchmark with similar experiences that your users play
Last year, we launched benchmarks to help you compare your experience with others based on our internal genre data. However, Roblox users often choose what experience to play across genres.
Today, we’re updating our benchmarks to be based on similar experiences that your users also play (with genre as a fallback).
We also use similar experiences (along with other signals) to decide what experiences are eligible to appear in recommendations, so these new benchmarks more closely reflect how our recommendations work.
To qualify for similar experience benchmarks, our model needs to find at least 50 similar experiences that have a large overlap in playtime and other signals with yours. You can see some of these experiences on your Experience Details page’s Recommended Experiences section:
Your similar experience benchmarks are updated daily. If you don’t qualify for similar experience benchmarks, you’ll still see benchmarks based on our internal genre data. We’ve added chart annotations to show when your experience transitions from genre to similar experience benchmarks:
New discovery documentation, explore mode, and a new experience overview
In addition to the changes above, we’re also excited to introduce:
- New discovery documentation that explains in simple terms how our recommendations and other discovery channels work.
- Explore mode to filter and view detailed breakdowns for any chart. To enter explore mode, click the expand button at the top right of any chart.
- A new experience overview page (in testing) that will help you better understand what metric to focus on at a glance. We’ll share more about that in the coming weeks.
As always, let us know in the replies if you have any questions or feedback!