Onboarding Redesign
Completion Flow
These people have been failed by the system
They've heard every promise before
Government, schools, the justice system — every institution has broken something. Now we're asking them to fill out a 7-step form before showing a single reason to trust us. Of course they leave.
The words we use make things worse
“AI-powered” reads as “a computer taking my job.” “State-funded program” reads as “the government is tracking me.” They're on a cracked phone with 30 seconds of attention and zero trust.
So we earn trust before asking for anything
The first 10 screens collect zero data. We lead with salaries, graduate stories, and free benefits. The form is last, not first.
Why these designs are mobile-first
Most applicants are likely accessing this on a phone — a prepaid Android on library Wi-Fi or limited data. The prototypes are designed mobile-first with responsive desktop scaling. One-thumb navigation, minimal data usage, offline-capable for the learning hub.
Responsive from 375px up
Fast on slow connections
One-thumb, one-tap
Service worker caching
Earn trust before asking for anything
Apps like Opal, Headspace, and Duolingo spend 10-20 screens building emotional investment before asking for $9.99/month. Emerge should do the same — except there's no paywall at the end. It's free. But they don't believe that yet, because every institution has broken promises to them before. So the first 10 screens collect zero data. They just build trust.
The emotional sequence
Each screen in the onboarding targets a specific emotion, in this order:
What I designed and why
Why “AI” never appears on screen
For the people Emerge serves, “AI” means job displacement — not help. They're reading “AI-powered” and hearing “a computer is taking my job.” They're seeing “automated system” and thinking “the government is tracking me.” Every piece of technology in this product is named after what it does for the user, never what it is technically. Here's what runs invisibly:
| Capability | What the student sees | What's actually happening |
|---|---|---|
| Benefits discovery | “You may qualify for Fair Fares, SNAP, IDNYC, free laptop” | Benefits Screening API + eligibility rules engine |
| At-risk detection | Coach texts: “Hey, checking in” | ML model tracking engagement → auto-alert coach |
| Smart nudges | “You're 80% through! 2 more modules.” | Automated triggers, personalized timing |
| Pre-fill | Fields auto-populate after “I receive NYC benefits” | ACCESS HRA Provider Portal cross-reference |
| Coach matching | Assigned to the right coach | Algorithm matching profile to availability + specialization |
APIs that power the experience
These directly power features in the prototype — the benefits discovery screen, the fast-track form, and barrier-removal workflows.
The shelter test
Every word must pass: would someone in a shelter, on a cracked phone, with 30 seconds of attention understand this?
- 6th grade reading level. No sentence longer than 15 words.
- Never say AI, workforce development, end-to-end, platform, or solution. Students see: “free job training.”
- Always say free, the dollar amount, the timeframe. “Free CDL training. $75K+ jobs. 4 months.”
- Name barriers and dismiss them. “Have a record? We work with that.” “No ID? We help you get one.”
- Use “you” and “your” constantly. Not “our students.” YOUR career. YOUR coach. YOUR future.
Search-optimized entry points
People Google “free CDL training NYC” and “jobs after prison New York.” Dedicated landing pages per trade + city, each with a local success story and one-click apply. SSR via Next.js for crawlability, structured data for rich results, and state-specific routing so New York applicants see New York programs immediately.