Keye
Subscription Marketplace — From three static screens to a company that later reached YC
01 — Snapshot
TL;DR
Product
A credit-based marketplace for flexible access to 150+ premium tools — Grammarly, Adobe, Crunchbase — without subscriptions. ClassPass for digital tools.
My role. Founding, and for most of Keye's life only, designer — designed every surface, wrote PRDs, ran research, hired the two designers who replaced me.
Timeline. 2 years on Keye (Jan 2021–Dec 2022): beta → credit system → marketplace → seed funding. What happened after I left is in the Pivot, below.
Impact
02 — The Problem
People were faking emails, sharing passwords, even paying for VPNs just to try a tool before committing to it
Users created fake emails for free trials, shared passwords in friend groups, even paid for a VPN for cheaper regional pricing — premium tools were priced for committed subscribers only.
"I keep creating new email accounts to get free trials. It's the only way to try things without committing."
Common — group chats across campus were already full of people splitting subscriptions and swapping logins; Keye gave that behavior a legitimate home.
ClassPass proved flexible multi-provider access worked; Apple One bundled someone else's services — neither filled the gap between free trial and full subscription. That gap was Keye's opportunity.
Friction ran both directions: users wanted to sample tools without committing, and lesser-known tools had no cheap way to reach new people. Keye's credit system gave both sides a lower-cost way in.
03 — Features
Three systems that carried the rest of the product
The Credit System
Five pricing models, ten months
Founders set the direction — monetize before runway ran out — I tested each model against real usage, treating price as a discovery lever, not just a revenue one.
Why a cap creates discovery, not just a limit
No monetization. The goal: prove people would use the product before asking them to pay.
Founders proposed a flat 10% commission, shipped before we had tolerance data. A follow-up survey later showed real tolerance was closer to 5–6% — that gap between charged and accepted shaped the cap that eventually shipped.
Closer to what users said they'd pay, but with no ceiling, heavy users consumed more value than the fee recovered — fixing that meant a cap, not a bigger fee.
Capping credits at 200/month turned scarcity into discovery: with a limited budget, users spent on products they'd otherwise have skipped for the brand they already knew.
The cap also capped the highest-intent users. Tiered plans let them buy up without loosening scarcity for everyone else — same mechanic, an escape valve for the edge case.
The Product Card
Five versions, one component
The product card is the surface every Keye user touches every visit — credits, waitlists, requests, referrals, mobile, the extension all live inside it. Each version added a funnel job: browse → status → merchandising → recovery → growth → cross-platform.
What the v2–v5 system converged on
Media, one fixed zone. A real screenshot or a brand-color lockup, sized to fit cleanly.
Favorite, top-right on the media. Lives outside the status/badge slot.
Title, always bold, first line below media. Holds more visual weight than the category.
Category, plain colored text. The pill shape stays reserved for the clickable filter chips above the grid.
Credit cost, bottom-right, same position every card. Price-informed before the click.
One badge slot: Leaving Soon, Coming Soon, Featured, and Waitlisted never stack — one lifecycle badge at a time, CTA color following it — green for buying access, blue for waitlisted, coral for capture-intent, black for a secondary action.
The Chrome Extension
A constraint the team couldn't engineer around — until the idea came from buying a rain jacket
Partner integrations were the plan until our co-founder left with the backend capacity. The idea came buying a rain jacket online — Honey auto-filled a discount code, and I realized an extension could do the job instead of partner engineering. Pitched Friday, built in two days.
The screen the extension made obsolete: copy a password here, paste it on the partner's site.
Install, prompted right on the grid — no separate onboarding flow needed.
Every active access, one click from Autofill — the same list, now in the browser toolbar.
The moment that mattered: the panel follows the user onto the partner's own site.
The version that shipped — personalized by name, accesses laid out as a grid.
04 — Evidence Over Instinct
The founders were moving fast on instinct. I slowed things down just enough to check if the instinct was right
My first task was the purchase flow: founders had baked in mandatory consent and a checkout survey dropping completion 30%. I moved consent to signup and the survey to a post-access popup, cutting drop-off 20% in one sprint.
Research base: early beta UPenn users, standout survey/support responses followed up via interviews, incentivized with Amazon gift cards.
05 — Tradeoffs & Hard Calls
None of these were mine to decide alone — engineering, marketing, founders, and investors each pushed back, and the product was better for it
Every hard call here was negotiated collaboratively — my job was bringing evidence into the room.
06 — Impact & Reflection
The most important thing a founding designer builds isn't the product. It's the foundation that lets the product grow without being rebuilt from scratch
What I got wrong. I thought unlimited credits would reveal honest user behavior — they didn't. With no cost to switching, users defaulted to the brand they trusted. The cap worked because removing that option forced a real tradeoff, finally showing what people would try when it cost them something.
Keye compressed the feedback loop from months to days, teaching me to treat every decision as a hypothesis — the untested ones were the ones that broke.
