interestingly enough, it’s all referred to as ads and they have actual visibility controls for me now
The curation teams needs to curate more games for todays picks 2.0 bro 
. These games are all good but come on there have to be more.
Like seriously if the nomination form is not enough you’ve gotta like post on social media or something cuz if this feature is here to stay id want it to be useful at least u know?
experiences which you dont promote because you prefer the same mindless brainrot slop games, Anyways (i wasn’t talking about today’s pick/standout games)
not a big fan how we dont even see the continue list on the mobile home page, and if im honest the only thing im really getting from this is “we changed today’s pick to standout games”
1 Like
Please bring back the “Because you liked game here” sort. I cannot stress enough how useful this was back in the day for this sort to randomly show me more showcases, or other games of the genre that were so niche and often without a playercount. It was the most useful discovery for finding games I enjoyed based off games I liked in the first place. The algorthym checked basic info such as name structure, type of genre, etc. and it was extremely beneficial to finding extraordinary games I would have never have found, ever.
1 Like
Is that not what the current “Recommended for you” thing on the homepage is?
No its not. It shows me active gameplay sessions and things that are trendy. It doesn’t do what this did. Recommended and Because you liked are two separate categories that have existed at the same time.
We’re adding a new signal to the Recommended For You algorithm: 7-day qualified play sessions per user. This signal measures the total number of qualified play sessions a user has in your game over the last 7 days. It helps capture whether a game retains users after the first session, leading them to return and engage meaningfully in the next 7 days and aligning with user retention on your game. As with other signals, you will be able to see metrics for this signal on your Creator Analytics dashboard under the Home recommendations tab of the Acquisition Page.
In response to community feedback, we introduced the ability for users to dismiss something they don’t like as a way to provide feedback on the recommendations they receive. This feature has been well-received, and we are now integrating this signal into our discovery system to further enhance the relevance of what users will see on Home. We will start testing and rolling out this integration over the next several weeks. You can start monitoring the stats on users who selected “not interested” for your game on the Creator Analytics dashboard under the Overview tab of the Acquisition Page.
Below we address a few additional questions you may have.
FAQs
Why are we adding more signals?
This is part of our ongoing effort to improve how our discovery algorithms support diverse games with varying characteristics and users with a wide range of interests.
Our approach is based on two key observations:
- Games are Different: The way we can tell users enjoy a game varies significantly across genres. Consequently, simple metrics like 5 minutes of engagement are insufficient to identify quality games across genres. For example, 5 mins could signal a user bounce in a deep-play game but peak success in a casual one. We require a large variety of user behavior signals to account for genre-specific differences.
- Users Behave Differently: User behavior changes based on context, such as device (e.g., phone vs. PC), time (e.g., weekday vs. weekend), and evolving interests.
By accurately learning what enjoyment looks like across different games and for different users, we can recommend the optimal set of games to maximize each user’s enjoyment, (which drives their long term retention on Roblox.
Why are we not sharing the weight of each signal relative to other signals? Or the thresholds within a signal. As a creator, how do I optimize my game?
The algorithm is moving toward dynamic and personalized weighting, meaning there will not be a single global formula in the future. Instead, the formula will be constantly evolving and unique for each user.
For creators, the core advice remains: focus on building a fun game. If your creator analytics show outliers where your game underperforms on a specific signal, consider that signal’s relevance to your genre. For example, a game meant for deep enjoyment might naturally have fewer repeat play sessions. If your game is underperforming on a signal that is highly relevant to your genre or audience behaviors, then consider ways to improve a player’s experience and fun.
Well seeing in the menu the latest played games is great but for it we got another [
] corporate renaming
With two successful tests now complete, we’re excited to roll out the following updates to Home.
-
Video Previews in Standout Games sort: Unlike static thumbnails, video gives users an authentic preview of real gameplay, and our test results show that this added context drives meaningful engagement - not just more clicks. The Standout Games sort will now primarily feature video previews, while we continue testing video across more sorts and surfaces. If you’d like to be considered for the upcoming tests, we encourage you to add an authentic video showcasing your gameplay to your game’s detail page. You can find guidance on dos and don’t for video in our documentation and check out examples by The Forge, 99 Nights In The Forest and Rivals.
-
Avatar sort on Home: We are rolling out a new Avatar sort on Home across mobile and desktop. This creates a prominent surface to reach users, enabling them to discover and purchase avatars directly from Home. The sort is fully personalized, including when the sort shows upt, its position, and the avatars shown.
a team has to monitor games abusing algorithm
Some games are currently abusing bot farms to gain an edge over other games, and the reason they can do it is because there wont be consequences or enforcement
Prioritize robux spenders, people who did age check, just anything that detects a human being
Design a system that detects and flags games that uses these inflated, monopolistic systems
Think it from a perspective of a new user
Do you want to divert them into slop kept up by bot farms?
Or to games that define roblox, games that people engineered as a game,
just something more healthy for both them and platform
we need bacons, but not too many
This update is significant for creators because 7-day qualified play sessions directly measures retention—arguably the most important metric for Home algorithm performance. Here’s how to optimize for it:Understanding the SignalThe algorithm now prioritizes games where players return within 7 days and have multiple meaningful sessions. A “qualified” session typically means 5+ minutes of engagement (Roblox’s standard threshold). This means a player jumping in for 2 minutes doesn’t count—you need sustained play.Actionable Steps1. Analyze your session data — Check Creator Analytics under Acquisition > Home Recommendations. If your 7-day qualified session count is low relative to your initial players, you have a retention problem, not a discovery problem.2. Focus on Day 1-3 engagement — Players decide whether to return in their first few sessions. Ensure your tutorial, core gameplay loop, and initial rewards are tight. If 60% of players don’t return for a second session, no algorithm tweak fixes that.3. Create return incentives — Daily rewards, seasonal content, or battle passes encourage Day 2+ engagement. Use DataStores to track player progression and trigger comeback notifications strategically (without overdoing it).4. Monitor session quality — Use RemoteEvents to log meaningful milestones (first boss defeated, level completed, etc.). Sessions with progression events likely count as “qualified.” Pad your stats with idle servers and you’ll tank your metrics.5. Cross-reference with your data — Compare your qualified sessions against total sessions in Creator Analytics. A 40% qualified rate is healthy; 15% suggests players are leaving before engagement starts.The Home algorithm now rewards durable games, not just novel ones. Build for retention first, discovery second.