kanchangaur.com / Designing for money you have already earned
LOS ANGELES

Designing for money you have already earned

2026-07-29 / essay / 3 min

Most payment design is about spending. Payouts are the opposite problem. The money is already the driver’s; we are only deciding how quickly and how clearly they can see it. That inversion changes almost every decision.

I work on payouts and on the taxes experience in the Lyft Driver App, embedded with the team that owns them. The two are the same surface at different time scales: what landed this week, and what it will mean next April.

##

Every number needs a receipt

The first thing I learned from driver interviews: nobody disputes a number they can decompose. Support tickets about earnings were almost never “this is wrong”. They were “I cannot tell where this came from”. A total with no breakdown reads as a claim. The same total with four labelled lines reads as a receipt.

So the breakdown is not a detail screen. It is the screen. Drag the week below and watch which numbers move, that legibility is the whole design.

INTERACTIVE / WEEKLY PAYOUT MON–SUN
$1,163.50 → bank by fri
RIDES THIS WEEK62
// fee only appears when the toggle is on, and it appears here, not at confirmation
##

The fee has to be next to the choice

Instant payout is a good product with a bad reputation, and the reputation comes from placement. When the fee is disclosed on a confirmation screen it reads as a catch. When it sits inside the toggle, before the tap, it reads as a price. Same number, same flow, different feeling.

BEFORE / AFTER, WHERE THE FEE SITS
BEFORE · fee disclosed later
Available now $248.10
Cash out
On the next screen Instant transfer fee −$1.24 · you receive $246.86
AFTER · fee inside the choice
Available now $248.10
Cash out $246.86
Nothing new appears after this tap.
// fig.1: the fee moved next to the choice and stopped being a surprise. Illustrative sheet, not Lyft’s: the fee shown is 0.5% with a $0.85 minimum
##

Taxes are a year-long feeling, not an April screen

Drivers do not experience taxes as a form. They experience it as a low hum of uncertainty about whether they have set enough aside. Our first attempt was a January summary; it was accurate and it was useless, because by January the decisions were already made.

The version that worked is a running estimate that lives inside the weekly payout, not in a tax tab. One sentence, updated every week, in the place drivers already look. The tax centre still exists for documents. It is no longer where the understanding happens.

> If a driver has to open a second screen to trust the first one, the first one is not finished.
##

What I would tell myself at the start

Read the support transcripts before the research deck. Design the failure states first, a delayed deposit is the only payout experience anyone remembers. And keep the tabular numerals; drivers scan these screens weekly, and digits that jump around cost more trust than any illustration can buy back.