BROOKLYN, NYC
SENIOR PRODUCT DESIGNER
21:21:30
Tuesday, August 18, 2026
Home Activation v3
Founding Product Designer
2026
The QUESTION
Dub is a brokerage where people invest alongside real investors’ portfolios. I owned the activation surface — everything between account creation and a first investment — including the problem framing and design.
Nobody asked me to redesign the home screen. I proposed it after noticing something in the funnel that didn't fit: users were stalling at the decision points, but the ones who got past them and viewed a few portfolio detail pages went on to copy at a healthy rate. The desire was there. We were just asking for it in the wrong order.
The home screen at the time was a funding checklist. "Welcome, Joy. Copy your first portfolio today." A progress tracker, a deposit button, three links. Every element measured our funnel rather than helping the person looking at it, and all of it was addressed to someone who had already decided to copy a portfolio. Most people opening dub for the first time have not decided that.
Here is what I wrote to the team when I pitched the work:
We're seeing a lot of stalling at key decision points, but once users do view a few PDPs they do end up copying. This is making me think that we may be "forcing" them to fund too early before they have intent.
Find out who's asking before you ask for money.
The PREMISE
Find out who's asking before you ask for money.
That has two halves, and I pitched both at once. The first is about sequence — the home screen's job is to build intent, not to collect a deposit from someone who hasn't got any yet.
The second half is the one that ended up mattering, and it was less comfortable to say out loud: dub is probably nobody's primary investing account, at least not at first. It's a satellite — somewhere people put a slice while the rest of their money sits with an institution they've used for years.
My hypothesis was that experienced investors would fund faster if we were upfront about that, not slower. And if it's true, then the users who bring real money aren't the ones a topline activation number is describing, so designing to that number will keep pointing us away from them.
I want to be precise about the order this happened in, because it changes what the results mean. I wrote that hypothesis down before we ran the test — not afterward, when a flat number needed an explanation.
The WORK
Move 1 — V2: trade the checklist for three things you can do without money
V2 (we called it micro-commitments) led with "Three ways to get oriented. Get comfortable with how investing on dub works — before putting money to work." Three cards, each a small reversible action: Follow a Creator (see how different strategies think and make decisions over time), Watch a portfolio (get real updates when a portfolio changes, without copying yet), and Ask dub AI (help making sense of what you're seeing).
The design choice underneath is that these are commitments, not education. A tutorial asks you to read. Following a creator asks you to pick one. That's a real decision at almost no cost, and it leaves the product knowing something about you it didn't know a second ago.
Funding and copying weren't removed. They moved to a persistent bar at the bottom — "Ready to start investing? Deposit" — available the entire time and demanded at no point. The tradeoff was explicit: we accepted a slower path to the first deposit in exchange for a deposit that meant something.
Move 2 — V3: stop orienting people and just ask them
V3 replaced V2's carousel with a question: "How do you want to invest?" Three routes underneath it — Let us find your match, Copy-trade real investors, and Build your own portfolio.
Three things converged to make this the right next move. The income result from the first test said that averaging our users together was hiding more than it revealed, so we should sort them at the door instead. Half the PM read on V2 was that we were getting too cute — orientation is a detour, and a detour is still a delay. And by this point we'd refined the AI matching, and other work had shown us something we hadn't designed for: a share of users want more control than copying gives them. They'd rather build or adjust a portfolio than mirror someone else's. So for the first time there were three real destinations worth routing someone to.
That last part matters more than it sounds. You can't ask "what do you want" until you have more than one answer to offer. The V3 home is only possible because the product underneath it stopped being a single funnel.
The OUTCOME
V2 — what the average was hiding
Among users reporting income above $100K, V2 lifted copy conversion by 30% — and it held regardless of investing experience, which was the part I didn't expect. Experience wasn't the variable. Having money already invested somewhere else was, exactly as I'd guessed before we ran anything.
Across all users the same screen came out flat. The distance between those two readings is the finding. A flat average was two populations pointing in opposite directions, and a single activation number treats our users as though they were one kind of person. I'd been reading a measurement problem as a design verdict.
V3 — asking directly
Arm | What's different | vs. control |
|---|---|---|
Control | V2 | — |
V3 Home | "How do you want to invest?" — match, copy-trade, or build your own | +45% |
The success metric changed between the two tests, deliberately. V3 routes people toward building and customising portfolios, and building a portfolio never produces a "copy" — measuring copy conversion would have undercounted the design by construction. Executed investment captures every path the screen offers. (It was also, in fairness, a different PM.)
One thing worth saying plainly: these were UX decisions as much as they were test outcomes. The experiments told us which direction to move; they didn't make the calls. V2 shipped because the structure was right, and it outlived its own topline — the Recommendation Engine UX is built on it. V3 shipped because asking someone what they came for is the right way to open a product, and because the PMs who ran these experiments and know this company read the results as enough to move. I agreed with them.
J.CHEN
Brooklyn, NY
joyhelenachen@gmail.com
Set in Sorts Mill Goudy and Satoshi. Built in Framer, 2026. The logotype is borrowed from J.Crew, with affection.
Ask for the Soundtrack: "Joy," composed for the proprietor by Steven Mudd, 2026.
J.CHEN
Brooklyn, NY
joyhelenachen@gmail.com
Set in Sorts Mill Goudy and Satoshi. Built in Framer, 2026. The logotype is borrowed from J.Crew, with affection.
Ask for the Soundtrack: "Joy," composed for the proprietor by Steven Mudd, 2026.
J.CHEN
Brooklyn, NY
joyhelenachen@gmail.com
Set in Sorts Mill Goudy and Satoshi. Built in Framer, 2026. The logotype is borrowed from J.Crew, with affection.
Ask for the Soundtrack: "Joy," composed for the proprietor by Steven Mudd, 2026.