Marsh
Mallow, two weeks ago you walked me through the plan to shift 'Recommended For You' to a 28-day retention window. What happened after that?
Mallow
A lot. Bottom line: it went live. Rollout began on June 15 — no longer a heads-up, it's running. And the same day, Roblox published not just the developer announcement but a platform-level newsroom post, 'Optimizing Discovery'. That's a company-level statement of intent.
Marsh
What's the difference between the dev-forum announcement and a newsroom post?
Mallow
Audience and intent. The forum is about implementation; the newsroom is about company strategy. This time Chief Growth Officer John Ciancutti fronts it: 'Great games should be easy to find. With millions of games and 132 million daily active users on Roblox, connecting players with games they'll love is more important than ever.' That's discovery being elevated to a top-line priority.
Marsh
So what's the goal, stated plainly? Did they actually say it?
Mallow
Unusually plainly. In Roblox's words, games that win attention with exciting thumbnails but don't deliver long-term value 'were displacing games that players value for the long term, actually harming long-term retention and engagement.' So the goal is killing clickbait. The change we discussed before — dropping QPTR, adding first-play bounce — all points at that one thing.
Marsh
Is it working?
Mallow
Roblox says yes. In testing, the new algorithm 'did a significantly better job surfacing games that retain players over time, leading to an increase in daily active users and engagement across the platform.' But no specific % is disclosed. So for now it's 'Roblox says it's working' — verify it against your own numbers rather than taking it on faith.
Marsh
You covered the mechanics last time (7→28 days, 3 buckets, first-play bounce). What's newly front-and-center now?
Mallow
Social. The signal that moved to the front is 'intentional co-play days' — days a player intentionally plays with friends: direct invites, joining a friend's session, private servers. Roblox is now explicitly weighting 'playing together,' one of the platform's strongest retention drivers.
Marsh
How does that change the design of a company game like ours?
Mallow
'Playable solo' stops being enough. You need to build in 'a reason to bring a friend and play together' — invite incentives, an experience that's clearly better/more fun with two or more, party and private-server flows. Treat co-play as a planning premise, not a bolt-on feature. That's the biggest practical impact this round.
Marsh
Honestly, I worry about new titles. If it watches 28 days, how does a just-launched game get its first exposure?
Mallow
That's the most-debated point right now. On the developer forum, people keep asking 'what about cold start for new games under a 28-day window?' and 'are Day 1 / 2–7 / 8–28 weighted equally?' — and Roblox hasn't answered clearly. A retention-first bar can raise the entry hurdle for newcomers. This is one to keep tracking.
Marsh
You mentioned there was trouble earlier, too?
Mallow
Yes. Before the live rollout, an April 10 RFY test had some games report Home impressions and concurrents (CCU) dropping 70–80% on the forum. Algorithm changes create big volatility while they're being tuned. So even after go-live, watching 'which bucket is leaking my traffic' every week is mandatory.
Marsh
As an operator, what do I change this week?
Mallow
Four things. (1) Add co-play flows — invite rewards, fun that assumes cooperation, party features. (2) Build Day 8–28 reasons to return — update cadence, live events, progression. (3) Keep thumbnails honest — deliver the icon's promise in the first session and lower first-play bounce. (4) Check Creator Analytics ⟩ Acquisition ⟩ Home Recommendations weekly. On top of last time's actions, this round adds a layer of co-play design.
Marsh
How should I read this alongside the previous article?
Mallow
The previous one (RFY revamped around 28-day retention) is the mechanics. This one is the two-weeks-later confirmation — live, the official goal, co-play moved to the front, and the open questions. Read together, you get both 'what changed' and 'what to do now.'
Marsh
Sum it up?
Mallow
(1) Live June 15. The newsroom post 'Optimizing Discovery' + the CGO quote spell out the goal: kill clickbait. (2) Roblox claims better surfacing and higher DAU/engagement in testing (no % disclosed). (3) The headline signal is intentional co-play. (4) Open: new-game cold start, and per-bucket weighting. (5) Operators: co-play flows, Day 8–28 return reasons, honest thumbnails, weekly monitoring.
Marsh
So instead of 'bait one click,' it's 'play with friends and still be here a month later.' The axis the algorithm rewards has finally lined up with the value we should be building anyway.
