CASE 03, CREDIT CARD CROSS-SELL · GEMINI · 2024

1,000 users
to 30,000.

THE MOVE

Strong product, weak distribution. I moved the cross-sell to peak intent: the final step of onboarding, after a user had already committed.

RESULT · USER GROWTH
30×

Credit card holders, post-launch. From a base of ~1,000 users to 30,000 in the new cross-sell flow. A distribution problem, solved.

BEFORE
1,000
30,000
AFTER
ROLE
ASSOC. DIRECTOR, GROWTH
TEAM
1 PM · 1 DESIGN · 4 ENG · 1 DATA
DURATION
2024
NORTH STAR
CARD ATTACH RATE → MTU
STATUS
SHIPPED · ITERATING
THE PROBLEM

Gemini’s crypto rewards credit card was a strong product on paper: real cashback, clear retention value, a natural fit for the existing user base. It was also invisible. Roughly 1,000 users had it.

The card lived outside the core trading experience. The cross-sell, when it ran, fired in the middle of a session, when the user’s mind was on a trade and the card looked like a distraction. This was not a product problem. It was a distribution and timing problem.

“The card wasn’t failing on its merits.
It was failing on its moment.”
, STRATEGY MEMO, WEEK ONE
THE BET

The thesis: a user who had just completed identity verification and funded their account had crossed the highest-friction barrier on the platform. They had demonstrated commitment. They had also just experienced the full weight of setup work, and were maximally receptive to additional value.

Move the cross-sell from mid-session to the final step of onboarding. Find peak intent. Test relentlessly into placement, framing, and conversion mechanics until we found the variant that compounded.

PLACEMENT, WHERE INTENT IS HIGHEST
Step 1
Sign up
low intent
Step 2
Identity
rising
Step 3
Fund account
committed
Step 4 ★
Cross-sell
peak intent
Step 5
First trade
activated
// the right offer at the right moment ↑ ANCHOR POINT
THE EXPERIMENT

I owned the cross-sell strategy and stayed hands-on with the experience. We ran A/B/n across three axes: placement (mid-session vs. end-of-onboarding), framing (rewards-first vs. utility-first vs. portfolio-first), and conversion mechanics (single-tap apply vs. educational lead-in).

The winning variant: end-of-onboarding placement, rewards-first framing, single-tap apply with a lightweight rewards calculator inline. Anything that read like an interruption lost. Anything that read like a finishing gift won.

THE OUTCOME

Card holders grew from ~1,000 to ~30,000, a 30× lift. The qualitative win was bigger than the quantitative one: card holders became more engaged platform users overall, lifting MTU and contributing to the broader north-star.

Strategically, this rewrote the role of onboarding. It stopped being a one-time setup task and became a multi-product adoption moment. That framing now anchors how we think about every new product Gemini launches into the existing user base.

NEXT

Referrals →