A percentage of every purchase — not the spare change.
Microinvesting that scales with what you spend — 1–10% of every purchase, not the loose change. Built solo in five weeks: research, brand, design system, and a working prototype on live financial infrastructure.
Every screen, designed.
Two onboarding flows, four product surfaces, and the stacking loop between them. One designer, five weeks.
Scroll →
Round-ups count your transactions. They don't measure your spending.
A round-up is whatever it takes to reach the next dollar — so what you invest depends on how many times you tap a card, not on what you tapped it for. Buy ten small things and you invest ten times. Buy one large thing and you invest once.
Two purchases, sixty-six dollars apart, under each mechanism:
| Purchase | Round-up | 1% | 5% default | 10% |
|---|---|---|---|---|
| $1.50 coffee | $0.50 | $0.01 | $0.08 | $0.15 |
| $99.50 groceries | $0.50 | $0.99 | $4.98 | $9.95 |
The round-up column is identical. Both purchases end in fifty cents, so both contribute the same — the dollars are discarded before the calculation starts. Every percentage column scales by 66×, exactly matching the 66× difference in price. At the low end round-ups win a cent or two; that's the honest trade for a mechanism that responds at all.
Both products have a range. Only one is in your control.
Acorns' range is $0.01–$0.99 per transaction — fixed in dollars. Wealthstack's is 1–10% — fixed in proportion. Put both in the same unit and the difference stops being a preference:
On a coffee, Acorns can take two-thirds of the purchase. On a grocery run, a hundredth of it. The same setting, a 6,500× swing in effective rate, and no way to choose where you land — the cents decide. The percentage band doesn't move, and picking it is the one decision the product asks for.
A $10.01 purchase invests 99¢. A $10.99 purchase invests 1¢.
Round-ups don't just ignore the price — they run backwards to it.The ceiling isn't a policy. It's arithmetic.
Every round-up falls between one cent and ninety-nine, and that range never moves. So the monthly total is really just transaction count × fifty cents — and count barely changes with income. Earning more doesn't make you buy more things, it makes you buy costlier ones. The mechanism goes quiet exactly as you start to succeed.
The fix
A share of every purchase instead of a share of every transaction. One number, 1–10%. What you invest becomes a chosen fraction of what you spend — so it grows when your life does.
Same transactions. Same spending. Only one of them moves.
Acorns publishes the benchmark: Round-Ups customers invest an average of $45 a month — about 90 transactions at a typical fifty-cent round-up. Hold those 90 constant, vary only what was spent, and put both ceilings side by side:
| Spent that month 90 transactions | Acorns round-ups | 5% $100 cap · default | 5% no cap | 10% no cap |
|---|---|---|---|---|
| $1,200 | $45 | $60 | $60 | $120 |
| $3,045 | $45 | $100 capped | $152 | $304 |
| $6,000 | $45 | $100 capped | $300 | $600 |
The Acorns column never moves. Five times the spending, same $45 — and no setting raises it, because the cents don't know about the dollars. Wealthstack ships with a ceiling too, at $100. The difference is that it's yours: raise it to $500, or turn it off and let the rate run.
$3,045 is my own spending, not an average. Default is 5% against a $100 cap; the cap is adjustable to $500 or removable entirely.
Both automate. Neither one watches what you spend.
Both run the same two mechanisms: round the change, or move a set dollar amount on a schedule.
| Capability | Acorns | Stash | Wealthstack |
|---|---|---|---|
| Invests without you acting | Yes | Yes | Yes |
| Amount responds to what you spend | No | No | Yes |
| You set a rate, not a dollar figure | A multiplier, not a rate | No | 1–10% |
| Contribution grows in a heavier month | No | No | Yes |
| Monthly ceiling you choose | No | Implicit | $100–$500, or none |
| Diversified default portfolio | Yes | Yes | Yes |
| Budgeting built in | Limited | Limited | Yes |
| Controls at first run | One — a multiplier | One | One — a rate |
Stash's Stock-Back pays up to 1% of a purchase in stock, which sounds like a percentage. Stash pays you a percentage; it doesn't invest a percentage of your money.
Answering the multiplier
Acorns lets you multiply round-ups by 2×, 3× or 10×, which looks like the answer to all of this. It isn't — a multiplier scales a number that was never connected to the purchase, so it multiplies the disconnection along with the amount.
| Same setting, two purchases | $1.50 coffee | $99.50 groceries |
|---|---|---|
| Acorns at 10× — invested | $5.00 | $5.00 |
| …as a share of the purchase | 333% | 5% |
| Wealthstack at 5% — invested | $0.08 | $4.98 |
| …as a share of the purchase | 5% | 5% |
At 10×, a coffee costs you three times its own price and a grocery run costs you five percent. There is no multiplier that fixes this, because a multiplier is a constant and a percentage is a ratio. The two only agree at a single purchase size: 10× equals 5% when a purchase is exactly $100, 3× equals it at $30, 1× at $10. Every other purchase — which is to say almost all of them — misses.
Which leaves the user with a setting they can't reason about. Choosing a multiplier well would mean knowing your own average transaction size and doing arithmetic on it. Choosing a percentage means answering one question: how much of what I spend do I want invested?
The person who won't start, and the person who outgrew the app that started them.
“I know a savings account isn't doing anything. I just don't know where to start.”
- Pain
- Investing is opaque. A CD is legible, so it wins by default.
- Goal
- Beat a savings account without learning a new vocabulary.
- Behavior
- Saves consistently. Has never opened a brokerage account.
“Round-ups were great when I was broke. I earn more now and it still puts in forty-something dollars.”
- Pain
- The mechanism doesn't scale, so the app quietly becomes irrelevant.
- Goal
- Put more in automatically, without turning investing into a hobby.
- Behavior
- Started on a round-up app, still has it installed, has stopped noticing it.
One number. It just happens to be the right one.
Acorns gives you one control too — a multiplier on the cents. This is one control on the dollars: 1–10% of every purchase, 5% by default. A cap sits behind it at $100 for anyone who wants a hard ceiling, movable to $500 or switched off entirely.
Seven screens to start stacking. No social security number.
Sign up, connect a card, set a rate, go. The regulatory burden is real — it just isn't required yet, because nothing has moved.
Scroll →
Connecting the card first is what makes the percentage answerable — without spending data, "5%" is a number nobody can picture.
The number on your dashboard hasn't moved any money.
A stack is an accounting entry — your spending earmarks an amount and the app tallies it. Nothing leaves your bank.
Two figures rather than one, because collapsing them implies money that isn't in the market. Nothing is held — so this sits outside money transmission, and verification can wait until money moves.
The edge case that shapes it
A tally can outrun the balance behind it: stack $37, spend down to $12, sweep fails. The app takes what's there and says so. It never overdrafts.
The hard questions arrive once — at the moment they're finally justified.
A first transfer means a brokerage account must exist: SSN, ID, selfie, employment, income, disclosures. Asked here, it answers its own "why do you need this." The same screens, moved, stop feeling like an interrogation.
Scroll →
Eight named steps, one question per screen, defaults preselected, and the verification wait broken into three named stages instead of a spinner.
Offence and defence, on the same four tabs.
Wealth is built by small daily decisions in two directions — money working, and money kept.
One control at first run, six available in Settings. Depth should be reachable, not mandatory.
Six of eight participants stalled at the same screen — and it wasn't the one I expected.
Eight moderated sessions on the working prototype, each walking the original onboarding order end to end: sign up, verify identity, set a rate, connect a card. I watched for hesitation, not just completion.
The SSN screen itself tested fine in isolation — participants understood what it was asking. What they couldn't answer was why it was first, before the app had shown them anything it actually did.
"I haven't even seen what this does yet and it wants my social."
P4, 29, tested the original onboarding orderThe screen order was the finding, not the screen
Every KYC field passed comprehension checks: participants knew why an investing app needs a legal name, a date of birth, an SSN. The friction wasn't the question, it was the sequence — the same eleven fields with nothing invested yet to justify them.
| Where the session stalled | Original order | Proposed order |
|---|---|---|
| SSN / identity questions | Screen 2 of onboarding | Deferred to first transfer |
| "What does 5% mean for me?" | Unanswerable — no card connected yet | Answered from real transactions |
| Time to first completed stack | Not reached by 3 of 8 participants | Reached by 8 of 8, in moderated retest |
A second round, moderated, walked five returning participants through the reordered flow — card first, rate second, verification deferred to the first transfer. All five reached a completed stack. None flagged the identity questions as a problem when they met them later, already inside a working product.
Eight participants, first round; five returning participants, second round. Recruited from people who described themselves as "meaning to start investing." Not a claim of statistical significance — a small, moderated read on where a first-time user's patience actually runs out.
I was asking for a social security number before anyone had seen the product.
Watching six of eight participants stall at the same screen, I went back through the built flow and marked every screen decided or accepted. The onboarding order was accepted — I'd never chosen it. If the argument against the category is controls you never use, then five setup controls behind twenty compliance screens is the same mistake.
Two setup screens become one. The rate is the decision; the cap sits underneath it as an optional guardrail rather than a second question. And because the card is now connected first, the screen can answer the thing nobody could previously picture — what 5% of last month actually was.
Designed, not yet built. The two-screen version above is what currently ships.
Three decisions, and what each cost
- Split onboarding in two
- Verification moves to the first transfer. Cost: profiles that stack but never verify, and a pre-verified state to design.
- Connect the card first
- The rate screen can say what 5% of last month was. Cost: bailing at the SSN traded for bailing at the bank login.
- Five controls down to one
- Only the rate changes what the product is; the cap is a guardrail with a working default. Cost: settings now live in two places.
Infrastructure first, so everything downstream moved faster.
Tokens, components, a validated chart palette, and a dark mode designed rather than inverted — built before the screens.
No tickers, no confetti, no red flashing numbers. The interface should feel like compound interest, not a casino.
- Plaid
- Read-only transaction access — what makes a percentage possible at all.
- Alpaca
- White-label brokerage: account opening, KYC, fractional execution.
- Tally, not custody
- An accounting entry until the sweep. No funds held, no money transmission.
- Sandbox build
- Production runs ~$1,000/month before marketing. Too costly to test a hypothesis with.
I built a product about restraint and couldn't apply it to my own scope.
Five weeks from nothing to a working prototype on live financial infrastructure.
Sector tilting is the clearest miss — a control I'd criticised the category for, shipped because building it was more fun than deciding against it. A feature earns its place if it changes what the product is, not what it can do.
What's next
Test the split onboarding against the old order. Put the cap-and-rate relationship on the setup screen. And find out what people think 5% of their spending costs them — right now, neither they nor I know.
Try Wealthstack
A working sandbox build — set a rate, connect a test account, watch a stack accumulate.
Join the TestFlight beta
