Journeys Scaling Plan
Controlled growth to 1M users without breaking the experience.
How we open Journeys to all Edge users while our backend catches up.
Miaomiao Lei · March 2026 · C2 Planning
The Situation
Two forces are about to collide in C2.
~1K
Active users today (behind Copilot Mode gate)
1M
Backend capacity ceiling (Picasso)
Apr 9
EN-all blog launch (visibility spike)
C2 S2
CM gate removed GA Journeys — Copilot Mode gate removedmdash; open to all Edge users
The tension: We want aggressive growth, but we can't let growth outrun capacity and degrade quality for everyone.
OUR PRINCIPLE
Every user who wants Journeys should be able to opt in. If we can't serve them yet, we tell them honestly and follow up when we can.
The User Journey
Three possible paths after opt-in. We never silently fail.
Three User-Facing States
What the user sees at each capacity level.
| State | NTP Experience | Settings | When |
|---|---|---|---|
| Activated | Journey cards appear immediately | Toggle ON · "Active" | 0 – 800K users |
| Setting Up | "Setting up your Journeys — cards will appear within 24 hours" | Toggle ON · "Setting up…" | 800K – 950K |
| Waitlisted | "You're on the list! We'll activate Journeys for you soon." | Toggle ON · "Waitlisted" | 950K – 1M |
UX Commitments
Non-negotiable principles for the throttled experience.
Toggle stays ON
The user expressed intent. We never silently flip it back to OFF.
Clear status everywhere
NTP card + Settings page both show what's happening. No mystery states.
Proactive notification
Waitlisted users get notified when activated. They don't have to keep checking.
Opt-out anytime
Waitlisted users can disable the toggle instantly, freeing their slot for others.
Growth Strategy
Growth doesn't stop — it modulates. Levers auto-adjust as we approach capacity.
Open Questions
Key unknowns we need to resolve before shipping.
How does the backend communicate capacity status to the client?
The client needs to know which state to show (Activated / Setting Up / Waitlisted). New API endpoint, piggyback on opt-in response, or config service?
How do we monitor backend capacity in real time?
Active user count, opt-in velocity, queue depth, P95 latency. Does Picasso expose these today, or do we need a new dashboard? What thresholds trigger tier transitions?
Can we surface activation notifications to waitlisted users?
Edge notification center, NTP card change, or push notification? What infra exists today?
What is the right tier threshold model?
Align with Growth checkpoints (100K / 250K / 500K) or capacity-based (800K / 950K / 1M)? Need alignment with Ben and Linjun.