Roblox’s Algorithm Changes Are Killing Our Game

[Du_cen]

It’s been nearly a year since the creation of this post. Although Roblox has continued to apply changes to their game recommendation algorithm in favor of newer games, they’ve removed the requirement to have other stats such as Session Time, Retention, and so on.

Now, all that seems to matter is the games thumbnail and the qualified-play through rate rather than the other KPIs. Roblox will give a game more impressions per day with this stat increased compared to a game that has amazing Average Session Time, Day 1 Retention, D7, Payer Conversion, and other metrics.

The image above is an example of a game that has stats below the benchmark but is still getting pushed by Roblox.

This change removes the entire process of game designing, and feeds players games that fall short in terms of their benchmarks.

Now, imagine having a game that has great stats but the qualified play through rate is around the 50th percentile in it’s genre. The current system is sabotaging experienced developers and gives chances to games that shouldn’t be given their impressions.

If you’re experiencing the same issue, please reply to this.

[Alkan]

It seems my game is being killed again.

Specifically, my game is no longer really getting Home Impressions.

My game, 8.5 years old with a dedicated playerbase(poor retention, good session time, and great monetization):

Compared to my newish mining game(good retention, good session time, bad monetization):

An average of 7,000 Home Recommendations for an established and successful game vs. 70,000(I ignored the recent upwards spike) Home Recommendations for a newish game that’s just a tech demo?

Observe the Home Recommendation impressions vs Search impressions as well. The first graph is completely unnatural compared to all other graphs I’ve seen; home recommendation should be much higher than search. Also something happened on September 3, my impressions got cut from 20k to 7k.

I’ve asked other developers, one of a competing RTS game and another of some quick clickbaity game with bad metrics; both had stupendously higher home impressions than my game, and their data looked normal; home making up the lion’s share, search way lower, and the other impressions at rock bottom.

And heres data from September 2023:.

My home impressions are down 93%, meanwhile Search is down 22%, for context

I can’t do anything with this level of impressions, it doesn’t add up to the size of my game.

Is there a hidden metric that’s being pushed now? How do I turn this around?

[Ovon30650]

Similar situation for me. Ive pushed away from updates and into basically baiting players to get a higher click rate to avoid being thrown out of recommended. Almost all popular ish games are now cycling their thumbnails most with low quality to stay on recommendations. This is a horrible system.

[RoamingDev (abr)]

Felt like sharing my recent issues with the algorithm.

We released a game during late august. Advertised for 3 days to test it’s initial stats.

After the test sponsor had finished, we optimised the game and sponsored it for 2 weeks with a reasonably high ad credit amount. Just as the sponsor was nearing it’s competition date, we started to see a rise in players coming from home recommendations, and a steady incline with our home impressions too.

We kept a steady CCU of ~300 players hitting ~600 on a peak day. This was purely organic. The sponsor had ended and we were running solely off our impressions from the home page and the search terms. Not even 48 hours after we felt like we had a steady player-base and could start working on another game to get a pipeline going, our game was struck down and our impressions were cut and continuing to be on a steady decline by the looks of it too.

We went from having a stable ~300 CCU to now only having ~50 CCU.
I guess the only thing I can do now is wait and hope the algorithm helps us out. Good luck to everyone else.

[photonavius (photon)]

My game also suddenly started decreasing (in the middle of sponsors) in impressions despite stats improving everyday:

right now my returning users are higher than my new users

Update, this is another game:

[Ovon30650]

I have compared my stats on my game Space Tycoon with another Tycoon (dragon merge) and have higher in almost all places including session time and monetization yet the game won’t be featured anymore? (because its older?)
Its unfair that the home algorithm is obviously reflecting HUGE upticks(like from youtube video spam with millions of views or a recent release) Seems its 100% features “quickly rushed experiences”

It was a very POOR idea to release this algorithm without the genre update and making adverting so costly in the same year.

If the genre update is a bust were all cooked.

[Alkan]

My 13 year old game is reaching breaking point now; I dont have enough players at night to facilitate lobby creation, and even at peak hours players sometimes have to wait to be able to play. Getting very few players a day compounds the issue of new player retention and player activity. This is basically a soft shadowban.

[Ovon30650]

Roblox is really limiting and CUTTING off search results over like 30? (If you didn’t know roblox there are more than 30 tycoons or simulators!!!) Are we just done lying now and outright killing small games? This search system is literally awful maybe stop making it worse or even pretend to care!!!
Roblox to care about the top 300 and not just 50 in each category: impossible

[PyroIdentity]

Had to bump mine, It happened around October 29. Game

Update: It’s now resolved but this is still a major concern.

[imonurspeeddial]

This just happened to me as well, it’s really getting frustrating watching the algorithm change so often. It gives me less of a sense of stability on Roblox. Tomorrow my game could be on the front page, or it could die off completely.

[FlyingFrets (daniel)]

For some reason the algorithm changes weekly? One week my game is absolutely popping off, then dying, then popping off, and now dying again. It’s unfair to everyone and the top Roblox games aren’t affected by this at all, it’s all the small developers, goes to show how much Roblox cares.

[imonurspeeddial]

Can we seriously stop doing this? Almost weekly or biweekly algorithm tweaks have been going on for over a year. I’d like to remind the algorithm team that changing the algorithm can put people out of jobs, and having seemingly A/B testing every week is not very reassuring that I’ll be able to put food on the table next month.

[SnowdustDev]

Absolutely dealing with the same issues mentioned above.

Just recently we’ve been pushing out consistent, high quality updates every 2 weeks for both gameplay and monetization, and we’ve had literally #1 on average session time for several weeks. Nothing changed. In fact – our position got worse after making both the game and its monetization substantially better, and despite our metrics being at all-time-highs.

Despite this, our new user impressions have rapidly and substantially dropped over the past few months. In particular, a change to the description on Nov 3-4 triggered a 50% drop in home recommendations that never recovered, even after the description was changed back to what it was before.

Hope this can be addressed.

[puinn]

This is happening to 2 devs i know personally, what in the world is going on with roblox right now.

[Du_cen]

I’m still experiencing a similar issue. My QPTR is above the benchmark but I’m losing home impressions rather than gaining?

My other metrics are pretty decent. But my impressions only show a steady sign of decreasing.

[Nogora22 (nogo)]

Can vouch here How does recommendation tab work? Does it actually work? Qualified Play Through Rate Help genuinely don’t know what happened, I used to be at least in the 0.90% range and now I can’t breach 0.20% with the same thumbnails after going through multiple revisions of Thumbnail Personalization, destroyed my game even with the little plays it was getting already

Please make it so we can choose where our Thumbnails are shown via the new Genre feature.

Example: I want my game’s Thumbnail to only be showed to SOCIAL Genre, so it will only be shown to players who PRIMARILY play SOCIAL Genre games, and ONLY if I select an option will it be shown to people OUTSIDE the SOCIAL Genre

[Maxocado_Avocado (Max)]

Not even kidding, I actually get the feeling that better stats = worse algorithm performance

My game averaged 150 CCU a few weeks ago, so I added more maps, balanced it, changed some UI and overall raised my avg playtime by 2 mins, and my D1 retention increased by a percentage point as well.

Guess what - the games CCU can barely surpass 90 on a weekend now. I used to think I knew how the system worked but I guess it’s completely random

[mr_smellyman]

Yeah, these changes totally killed my game, especially after the “low quality filtering” update to home recommendations. We went from 100k impressions and 3k plays a day to 10k impressions and 100 plays a day. Nobody’s even playing it now, and we had amazing session times and monetization before.

[Ovon30650]

Same exact thing for me. Is ANYONE gonna talk about this? I had a great November/start of December then those changes roll in and everyone loses all their players. COME ON!? Were all just gonna let this slide?

[Ovon30650]

So I saw it mentioned once in one of these posts BUT WE NEED MORE COMBINATIONS FOR RECOMMENDATIONS.

Day 1 retention and retention CAN NOT BE WEIGHTED THE HEAVIEST AND TRENDS CAN NOT BE 1000X. It creates a hollow platform. “Some games are meant to be played in detail 1-2 times” not for 5 minutes 10 times.

But ROBLOX WEIGHTS IS AS: You played 2 times for 5 minutes much better than once for 20!!! Can we FIX THAT PLEASE!!! Some games are not as repayable… Just code more recommendations options. Or don’t and take 10 more months to do anything…