Test Drive
HealthPowr connected underserved New Yorkers to housing, healthcare, food, and employment resources through verified community-based organizations. The product worked. But the first thing users encountered wasn't any of that, it was a sign-up wall — create an account, verify your email — before experiencing any of it.
For someone navigating a housing crisis or a food emergency, uncertainty around a product's value = certain drop-off. I redesigned the flow to deliver value before asking for commitment. The result: completion intent increased by +153%.
Here's how.
Context
On paper, it seems like what you need. The salesman asks if you'd like to start the paperwork — but you hesitate. You haven't sat in it. You don't know how the car REALLY feels.
Now he hands you the keys. You pull out, feel the acceleration. You realize — hey, I really do like the way this car drives. The paperwork doesn't feel like a barrier anymore. It feels like the next logical step.
HealthPowr's original flow asked users to fill out the paperwork first. The redesign handed them the keys.
Problem
The product worked.
The entry point didn't.
HealthPowr's landing page had a conversion problem nobody had named yet.
Before truly understanding the product and if it will benefit them, users were asked to create an account, verify email address. For someone navigating a housing crisis or a food emergency, uncertainty was a reason to leave.
The question I kept asking wasn't how to make the sign-up form smoother. It was whether first-time users were seeing enough value to justify the commitment at all.
Research
No time for interviews.
So I looked outward.
Baymard Institute research identified forced account creation as one of the highest drop-off points — friction applied at the exact moment of highest intent.
The principle was clear, but what did the solution looked like in practice?
Researching into hero design patterns, I came across an article documenting a shift away from full-width lifestyle heroes. The most interesting alternative discussed was a so called "search-first" model — Opendoor being the clearest example. Left: You type your address, you get an offer. Right: Visual, mental model. The account comes later.
The logic is simple: let users experience the product's core value — in Opendoor's case, an instant home offer — before asking them to invest time or personal information. By the time the account creation step arrives, users already know it's worth it.
That observation produced one hypothesis: if users encounter HealthPowr's value before the login wall, it could significantly increase product flow completion and decrease the potential of drop-off.
Design Process
Start their search first.
Sign-up form last.
What if we moved the core flow of the product 50% in front of the sign-up wall? The login wall hits when the user tries to submit, the moment they're most motivated to finish what they started.
Communicate with real community resources in minutes and get personalized
help — no cost, no pressure.
The first two headlines talked about the platform — its simplicity, its proximity. The third talked to the user.
"Support is out there" is calm. It validates what the user already suspects before they've seen a single result. It's not a product claim — it's a statement of fact that the platform then delivers on. "Let's find it together" makes HealthPowr a partner, not a directory. The burden shifts. The user isn't searching alone anymore.
The period after "out there" creates a beat. A moment of reassurance before the ask.
Accompanying the header and search cta needed to be a visual, a proactive one. I used that space to handle the first objection — "okay but how does this actually work?"
With Claude I was able to rapidly design, prototype, and iterate on a looped, simplified animation of the product flow, layered on top of a labeled map design, as the flanking visual.
This approach does two things: it shows the product working, and it makes an abstract platform feel immediate and real.
Convo w Claude ?
Design Solution
1 — Landing Hero Redesign
2 — Services Bottom Sheet
3 — Account Creation Modal
Micro-interaction visuals ??
Testing
I ran a Lyssna preference test with 30 participants across two versions of the flow — original login wall, redesigned search-first — with one task: find food assistance near you in New York City. No hints. No guidance. Just the flow.
Before/ after Visuals ??
Results
Qualitative
"Because it immediately made you login to something, unsure if it was even the right place."
"It actually shows me what's available vs having to login or create an account first."
"I wouldn't really want to log into something to just see where I can get help."
??
Lessons
On AI-assisted workflows
I used Claude throughout this project to compress the distance between a decision and its consequence: framing a product problem, working through the logic, and pressure-testing a prototype in the same session. That speed changed what I was willing to question. The handoff artifacts reflected how the workflow matured: starting with HTML prototypes, expanding into TSX components, then full build packages with accent maps, CSS files, and implementation docs.
On staying productive when blocked
Working with a dev team that moved slower than the design meant learning how to stay productive in the gaps. The instinct is to wait. The better move is to treat every blocked moment as a design opportunity, to go deeper on the thing already handed off, spec the next surface before it's asked for.
On going wide
Five surfaces in three weeks means every surface gets attention and no surface gets enough. The lesson wasn't about working faster. It was about knowing which decisions deserved depth and which deserved a strong default, and making that call early.