Designing for a Platform Shift Everyone Ignored
Perfect Morning: a calm, 10-minute, once-a-day Roblox game.
The platform changed what it paid for, and almost no one redesigned around it. When Roblox's Creator Rewards began paying for daily return and audience growth instead of raw session length, I built the one thing the whole platform was optimizing against: a game that is intentionally short, calm, and playable once per day. I wrote the gap analysis, an enterprise-grade PRD, and the technical architecture, then designed and shipped it solo in 10 days. The constraint was not a limitation; it was the product.
KPI figures on this page are pre-launch targets, not measured results, and visuals tagged Illustrative are recreated to show the approach. The strategy, PRD, design rules, and architecture are real.
A platform changed its incentives, and the market did not move
Roblox's Creator Rewards pays developers for driving daily engagement and expanding the platform's audience. It replaced older systems like Engagement-Based Payouts, so instead of being paid mainly for total session length, creators now earn a baseline in Robux plus a share of platform purchases from the new and returning players their game brings in. In practice, the algorithm started rewarding 24-to-48-hour return rate over raw time-in-game, which quietly changed what a "good" game is worth building.
Almost everyone kept optimizing for the old metric: longer sessions, more grind, more time sunk. No one was designing for the new one. There was no game built to be intentionally short, once-per-day, and calm, even though that is precisely the behavior the new system pays for. The mismatch between what the platform now rewards and what the market still builds was the entire opportunity.
Before · Engagement-Based Payouts
Reward total session length
Optimize for time-in-game. Build grind, sinks, and long sessions. Everyone competes on the same axis.
Now · Creator Rewards
Reward 24-48h return + audience growth
Baseline Robux plus a share of purchases from new and returning players. The win condition is daily return, not session length.
Turn the platform insight into a product, on paper, before building
I held myself to an enterprise PM bar: a complete PRD before any code. Pick one user and prove why she returns daily, make the constraint the differentiator, size the retention mechanic to the metric the platform now pays for, design monetization that does not betray a calm product, and pre-commit to a kill gate. Only then, build.
Read the platform
Design for the new incentive, not the old one.
Constraint as product
Short, once-daily, calm, by design.
Monetize ethically
Hard rules that protect the ritual.
Gate the effort
A pre-set rule to stop and redesign.
From a platform incentive to a single, defensible concept
I paired the Creator Rewards shift with a scan of the calm-habit category outside Roblox, where daily-return behavior is already well understood: the streak loop in Duolingo, the gentle daily check-in of Finch, the ritual framing of Headspace. None of that thinking existed on Roblox. The concept chose itself: a deliberately short, once-per-day, calm game, because it is exactly what the new algorithm pays for and exactly what no competitor was building.
Platform signal
Payout now follows daily return, so a once-a-day ritual is aligned with the money, not against it.
Category proof
Calm-habit apps already prove people return daily for a short, low-pressure moment.
The empty quadrant
No Roblox game was short, once-daily, and calm. The gap was the whole reason to build.
One sharp persona, and the job she hires the product to do
I built the PRD around a single, specific persona rather than a vague "everyone," modeled on real behavioral patterns in the calm-app category. Designing for one person with a real job-to-be-done is what keeps a product opinionated.
Maya
21 · nursing student
Overloaded, time-poor, and wary of anything that wants more of her attention. She does not want another game to sink hours into; she wants one calm thing that asks for ten minutes and then lets her go.
Jobs to be done
- wb_twilight"Give me a small, calm ritual to start my day, without guilt or grind."
- check_circle"Let me finish and feel done, not pulled back in by a timer or a reward I will miss."
- local_fire_department"Give me a reason to come back tomorrow that respects my time."
The limit is the feature: 10 minutes, once a day
The core decision was a server-enforced 10-minute hard cap and a one-session-per-day rule. Not a suggestion, not a soft nudge, an enforced boundary that makes the product trustworthy: Maya knows it will never try to keep her longer than it promised. Enforcing it server-side is what makes the constraint real rather than cosmetic.
Locked
waiting for the day's window
Active
10-min cap, server timed
Complete
streak credited, journal saved
Locked till tomorrow
one session per day
A 30-day streak ladder, sized to the metric that pays
IllustrativeBecause the platform now pays for return, retention design is the business model. I built a 30-day unlock ladder with calm, earned milestones rather than punishing streak-loss mechanics, engineered toward a day-1 return target of 40%. The reward for coming back is a gentle sense of progress, not a fear of losing what you built.
Day 1
first ritual, first unlock
Day 7
new calm theme
Day 14
ritual customization
Day 30
milestone keepsake
Design target: D1 return >= 40%. The ladder's early rungs are deliberately close together so the day-2 return has an immediate, visible payoff.
Ethical monetization, written as hard rules
A calm product dies the moment monetization turns predatory. So I wrote the constraints into the PRD as non-negotiable rules, designed around dignity rather than whale extraction, because for this audience trust is the retention mechanic.
No loot boxes
No randomized purchases, ever.
No FOMO countdowns
No manufactured urgency or expiring deals.
Max 2 prompts per session
And never during the ritual itself.
Dignity pricing
Priced for fairness, not for extracting whales.
A fail-fast gate, set before launch
If D1 return is under 25% by day 14, stop and redesign
I committed to the kill threshold before launch, when I could still be objective. If the core return behavior was not there in two weeks, the answer was not to add features and hope; it was to stop, diagnose the funnel, and redesign. A pre-set gate is what turns "I am attached to this" into an evidence-based decision.
Then I specced the hard parts and built it
The once-a-day promise only works if the engineering is honest about time and privacy. I designed the architecture for the edge cases that quietly break habit products, then built the whole thing solo in 10 days.
Session-gating state machine
The single source of truth for locked, active, complete, and locked-until-tomorrow, so the cap and the daily rule cannot be bypassed.
Server-authoritative time
"Once per day" is decided on the server, with explicit handling for DST changes and players who travel across time zones, the edge cases that silently double or deny sessions.
Privacy-first journal
A reflective journal designed privacy-first, with text moderation, because a calm space has to be a safe one.
Risk register: I documented the failure modes up front, from DataStore throttling under load to a monetization prompt breaking the calm tone, each with a mitigation, so a bad week would be a plan rather than a scramble.
A platform thesis, turned into a shipped product
1 shift
platform incentive, designed around
Full PRD
before any code
10 days
designed & shipped solo
The real result is the product judgment: I read a platform-level incentive shift before the market did, turned it into a differentiated concept built on a constraint everyone else avoided, specified it with the rigor of an enterprise PRD, set an ethical monetization model and a fail-fast gate, and shipped it solo in 10 days. The D1 numbers here are targets, not a scoreboard. The transferable outcome is a product I could defend strategically at every layer, from platform economics down to the time-zone edge cases, which is the same judgment a TPM or PM brings to a roadmap.
What this proved about product strategy
Read the platform's incentives
The biggest opportunities hide in what a platform just started paying for and the market has not noticed.
The constraint is the strategy
A sharp limit that competitors avoid is a moat. Differentiation is often a deliberate "no."
Design ethics in, not on
For a trust-based product, the monetization rules are a feature, written into the spec, not an afterthought.