Roblox’s Algorithm Changes Are Killing Our Game

[spacemax_rblx (Max)]

Experienced this myself with a new game I’ve launched recently + a few of my other projects as well. Released a new game, it was growing more and more every day and even reached 660 CCU - and then the next day it dropped to ~200 and it’s been decreasing in terms of visits, CCU and revenue ever since.

No complaints from players, we didn’t release any updates/do any changes either. The game’s fall seems to have happened at the same time as the touched events breaking but I’m not sure if that’s just a coincidence.

What I’ve noticed is that Roblox has been making these algorithm changes multiple times over the last year. One of my games randomly spiked from 80 CCU to 500 about 5 times already, it’s almost like it’s in “waves”???

[samuelagent]

I have not experienced this problem, all of my games have been at stable player counts or rising. I don’t think having a general downward trend over all of your games is indicative of an algorithmic problem, rather it could definitely be due to user-related changes recently

[ruddev_ethan (gamer961)]

I think I’ve provided enough evidence that there is something funky going on with the algorithm, especially in combination with others finding the exact same things are happening with their games in the same time periods. Competitiveness is not to blame for us suddenly being shown to millions less people after years of built up consistency.

[Chemiclast (industrialCentrifuge)]

I second this, Home acquisition was a huge boon for me for a very short period of time (nearly 350k in one day!) but for the last couple of months it’s been on a steady decline

I understand that driving users via more controllable means is the ideal solution but the added handicap of stunted home acquisition isn’t helpful… I wish the algorithm wasn’t such a black box sometimes. It’s impossible to understand what’s happening with home page acquisitions

[samuelagent]

I have also experienced sudden drops in acquisition similar to your examples, I’m not disputing the idea that there have been changes to the algorithm in regards to your experience, I’m just saying that the changes that are made can’t be umbrella’d as being a “bug” or “error” or “bad decision” on roblox’s end to lower the amount of outreach it provides your game. It’s impossible for roblox to always maintain or increase the outreach of every experience forever. A drop in outreach can surely kill your game, but a drop in outreach is not always due to some inescapable error.

I have also lost high CCU games, games that reached 3k CCU in the past lost algorithmic support and nowadays average 100 CCU. I’m not sure if it’s fair to blame roblox for this, these kind of events are commonplace in every area of the game development industry

[porttoport (port)]

This is reality and is happening right now to many games.

I am not a holder of any major experience, but talking from players’ POV, discovery of new or entertaining games is very messed up. The same experiences keep being promoted in multiple sections on the discovery page, and no other experiences are. Examples of these games are Brookhaven, Blox Fruits, Pet Sim X, etc. (No hate to those experienced whatsoever, they worked hard to be where they are, but not in 3-4 categories at the same time)

And on top of that, any popular, more unique/startup experiences get hidden by mass-produced, low-quality experienced like obbies and “Escape the [something]” obbies. Those are experiences too yes, but most of them are repetitive, not unique, and usually from the same people, trying to cash-grab.

We need new and upcoming games to be featured more, perhaps a user-selected category, and for god’s sake, we NEED genres back! They’ve been promised a bit ago, where are they?

To conclude, the discovery page and discovery algorithm need rework and improvement ASAP.

Examples of bad discovery sorting:

Not anything major, but remember these two experiences

First experience is seen in the “Most Engaging” category, a misleading one.

Have you seen any of these experiences before?

How about now?

You must see them now…

Again, no hate to the above-shown experiences, they all worked VERY hard to get where they are, but perhaps shine to spotlight to some other experiences too…?

[ruddev_ethan (gamer961)]

A drop in outreach can surely kill your game, but a drop in outreach is not always due to some inescapable error.

I don’t know what to respond here other than I am going to bat for my developers since this is our livelihoods. We have remained on Roblox because it simply was the most stable place to run indie live service games. It made the most business sense.

If Roblox can no longer offer that stability, we are better off elsewhere. And this is all fully within Roblox’s control - algorithmic changes should be notified, statistics that it depends on should be made readily available for developers, all of this is part of Roblox’s developer relations and if they’re willing to nuke games off the platform in the course of a month, they cannot possibly expect us to keep our business here.

Lastly, the fact that our impressions stats have not reduced to the same level as our home-page new user acquisition stats points to me that they have blatantly started showing our game to the wrong people. Their algorithm is calculating interest incorrectly.

[OutlookG (Outlook)]

This isn’t really valid for us if we have significantly better stats across all metrics that matter for the algorithm (playtime, monetization, retention). We’re the game with the highest stats out of any game in our genre with our size, yet we’re being punished for no clear reason when we should be promoted more then any other game.

Its clear there’s a bug here if we aren’t being promoted even though we have better stats then everyone else. This doesn’t make sense from a business perspective or helps Roblox in anyway.

[CashyMone_y]

Gotta love that in the insights tab it tells me that my session time is good (top 7%), yet it keeps punishing the game lol!

[Noble_Draconian]

@samuelagent

I have not experienced this problem, all of my games have been at stable player counts or rising. I don’t think having a general downward trend over all of your games is indicative of an algorithmic problem

Just because you are not personally getting affected by a change does not mean the change did not occur. There is plenty of evidence that shows it was a problematic algorithm change.

[Solkryn (Soleste)]

Such a disgusting algorithm to punish newer or less popular games which could provide fresh new content to play instead of the same old boring front page stuff.

Sad to see you guy’s games get killed off by a bad algorithm,
Actually working on a game myself that is close to releasing so seeing this makes me hesitant to even put up a sponsored ad for it.

The fact that roblox recommends the same games over and over is clear that the current system is heavily flawed and favors the popular experience over the less popular.

Hoping to see this fixed as a majority of front page content are obbies and crappy simulators.
No offense to people who make those, just dissapointing we see so many of them and less of the more unique games.

[Solkryn (Soleste)]

This exactly, just because one person doesn’t experience the issue doesn’t mean it never existed / doesn’t exist.

Simply the fact that this can happen to ANY person or team making a game is ridiculous and ROBLOX should get onto it as quick as possible if they want new developers to work on their platform.

If we have to start making guesswork out of what games would even show up why are we pouring money into advertising our experiences?

EDIT: Before anyone comments, I know UGC is a huge focus for ROBLOX as of currently, however I think they should have more focus on fixing existing systems rather than making new ones accessible in experimentation.

[tu35d4y5]

I know this is going to be a very unpopular opinion, but you shouldn’t depend on the algorithm for the success of your game. You need to invest in other ways to promote your game whether it’s advertising or social or word of mouth or something else so that when algorithm changes do happen you still have several other good sources of traffic.

[Maxwell_Edison]

This is not just an unpopular opinion, but an irrelevant opinion.

What matters is consistency - Roblox’s algorithm showing signs of decay is bad for players, not just the developers. The discovery algorithm failing to show games that it normally would is inherently worse for players, who, y’know - rely on these discovery features to find games as much as, if not more, than the developers rely on them to get the players. It is a mutual exchange.

There’s also a good chance that even with heavy sponsorship reinvestment, it won’t fully recover back whatever changes Roblox made lost.

Whether or not you think developers should rely on these algorithms is inconsequential to the fact that if these algorithms for discovery are messing up somehow, it’s bad for everybody on roblox.

Coming even from a purely Player-oriented mindset it’s easy to see why this is an issue. Saying “Just don’t rely on the algorithm!” misses the point entirely.

[tu35d4y5]

I was going to reply but I’ll just say good luck with your game.

[KardashianKlan (Jezzie)]

Just wanted to report that my game has been experiencing the exact same thing as everyone here. It’s not just a coincidence I fear.

[qChronos]

This is probably because of the removal of the friend activity tab, I used to see bad business on my friend activity tab a lot but now I don’t (because its gone lol)

[filiptibell]

I usually don’t post about our games on here but discovery definitely feels like it has been getting worse, and much more so recently. We released a new game two weeks ago - Zombie Outbreak Tycoon - and our last 7 days show stats like these:

• 20+ minutes session length average across all platforms
  • Roblox sets a benchmark here at 7-14 minutes
• 23% D1 retention, 4% D7 retention, 20%+ W1 retention, W1 still stabilizing
  • Benchmarks are at 8-15%, 1-3% and 6-13% respectively

Let’s compare this to a different game we have in the same genre - Tropical Resort Tycoon:

• 15 minutes session length average across all platforms
• 19% D1 retention, 3.2% D7 retention, 14% W1 retention

So, what’s wrong here…?

The second game has been out for 3 years, had no updates for a full year, peaked at 32 thousand concurrent players with several thousand concurrent still, while the newly released game is seemingly hard capped at a few hundred concurrent players.

Pouring hundreds of thousands of robux into sponsoring is the only thing that seems to permanently raise the concurrent player count. The new game is also impossible to find through search & discovery, or in any of the personalized sorts, even though my recent play history is more or less completely made up of similar games and games of the same genre.

Meanwhile, games that released during a similar time period and that have had popular creators on platforms such as YouTube make videos on them seem to be doing much better, meaning external discovery is still doing fine.

Something is going on with discovery on Roblox, and I think it’s a lot bigger than just strange declines in player counts.

[Mimi_Dev]

Oddly enough I saw the exact opposite of this. During April we weren’t seeing much homepage acquisition and our game was struggling to attract players. However during may we suddenly saw an increase in home page acquisition and we thought it was recovering. But now in June we have dropped back to the April home page impressions numbers.

This is really frustrating as during May, the players we got were actually our target audience and interested in the game, so much so that we saw a recovery in our thumbs up-down ratio (that had previously been effected by our game being promoted to the wrong audience early release).

I wish we had more stability; its really difficult to update a game based on data that keeps changing. It’s hard to tell when we are doing something “right” by the algorithm’s definition.

[qChronos]

@OneEDM

Oh yes seeing what premium players is totally very useful.

I’m pretty sure that sort was made to highlight games with good premium perks, it had a lot of potential but ofc they ruined it