Three minutes after downloading Benjamin, we asked people to link their bank card. Most of them didn’t.

Benjamin is a US cashback app: link a card once, shop as usual, and earn rewards from local businesses automatically. No coupons, no receipts. But the whole product rested on that one step. Without a linked card, there was nothing to earn.

This is how I redesigned that step, without making it optional, and how card-link completion went from 30% to 90% in the week after launch.

I joined Benjamin as its only designer, and we launched the first version in January 2023. Card-linking sat inside sign-up: create an account, link a card, start earning.

Early data showed that only 30% of people who reached the linking step finished it.

For most apps, a weak onboarding step means lower activation. For Benjamin, it meant the product didn’t work at all for seven out of ten people. They had an account, but they could never earn a reward.

Getting more people through the form was the obvious goal, but it wasn’t the only one. A linked card is only useful if it’s the right card. I measured the work against:

• Completion: More people finish linking a card during sign-up.
• The right card: People link an account that can actually earn rewards.

Before touching the design, I wanted to know why. I interviewed [Edit: number] new users who had abandoned linking and reviewed [Edit: number] session recordings from [Edit: the funnel window, e.g. the previous two weeks]. [Edit: who ran the interviews and recordings review, and your part in it, e.g. “I led the interviews with our PM taking notes.”]

The same three issues came up again and again:

• People didn’t understand the ask. We were asking for bank details within three minutes of download, with almost no explanation of why, and no visible reason to believe it was safe.
• People didn’t know where to find their credentials. They had to leave Benjamin to look inside their banking app. [Edit: what the recordings showed happened next.]
• People were linking the wrong account. A debit card tied to a credit account couldn’t earn rewards, but nothing in the flow said so. They finished onboarding feeling confident and earned nothing.

[Edit: one anonymised quote or observation from the interviews or recordings.]

The form had usability friction, but the deeper failure was trust: people did not understand why we needed this information, whether it was safe, or which account would actually make rewards work.

The original flow: the ask came with no context

My first instinct, and the team’s, was to make card-linking optional. Let people explore Benjamin first, get comfortable, and ask them to link a card later.

We prototyped it, and it tested well in sessions. People liked not being asked for their bank details straight away.

But the data told a different story: people who skipped card-linking almost never came back to complete it. Optional meant abandoned. [Edit: how was this tested, e.g. a release to some users, and what share of skippers came back?]

That was the hardest call. Removing the requirement would have made sign-up feel better and the product work worse. So I kept the step mandatory and set out to make it worth the ask.

Instead of softening the requirement, I rebuilt the flow around one principle:

Explain before you ask, guide while you ask, reassure throughout.

Explain before you ask

Before the form, I added a screen written in plain language, not legal copy. It explained what card-linking was, why it was safe, and exactly what people would get from it.

That one screen changed the framing from “give us your bank details” to “here’s how to unlock your rewards.”

Guide while you ask

The recordings had shown two places people got lost, so the form addressed both.

First, it showed people where to find their credentials in their banking app, so they didn’t have to leave Benjamin to go looking.

Second, and less visible, it showed them which card to link: a debit card paired with a checking account. A clear illustration and one line of explanation told them why it mattered. Linking the wrong account was the difference between earning rewards and never seeing one.

Reassure throughout

Finally, I added a way to reach support from inside the flow. If someone got stuck, they could reach a person without leaving the screen, and without abandoning sign-up.

Help without leaving the flow

The redesigned flow

Who made it happen

I was the only designer, but this wasn’t a solo change. [Edit: how you got the team behind keeping the step mandatory after optional linking had tested well.]

• Analytics: [Edit: who owned the funnel data and the 30% / 90% numbers.]
• Research: [Edit: who ran or joined the interviews.]
• Engineering and the bank-linking provider: [Edit: which provider, and what it or engineering limited in the flow.]
• Security language: [Edit: who reviewed and approved what the explainer could say about safety.]
• Support: [Edit: who answered in-flow requests, and how.]

The people who weren’t ready. A mandatory step means anyone not willing to link a card on day one doesn’t get in. I accepted losing them because an account without a card could never earn. [Edit: confirm, and add what share still left at this step.]

A longer sign-up. The explainer screen and guidance add time before someone reaches the app. I chose a slower sign-up that works over a faster one that doesn’t. [Edit: confirm, or replace with the trade-off you actually weighed.]

Support on call. An in-flow support entry point only works if someone answers it. [Edit: who handled it, and what it cost the team.]

Card-link completion among people who reached the linking step increased from 30% in [Edit: baseline period] to 90% in the seven days after launch, measured in our analytics.

That seven-day cohort was [Edit: number] people. [Edit: if that’s small, say so plainly, e.g. “It’s a small sample, so I treat 90% as a strong early signal rather than a settled rate.”]

[Edit: what else changed in the same period, if anything: bank-linking provider, eligibility rules, where new users came from, other onboarding changes. If nothing did, say so in one sentence. If something did, disclose it here and soften the claim.]

The fields people filled in barely changed. What was missing was the context around them: the why that makes someone decide an ask is worth it.

Making the step optional would have been the easy answer, and it tested well. But people saying they liked it and people actually coming back were two different things. I learned to trust what people did over what they said.

If I could go back, I’d have pushed harder for card-linking education from the start. We left value on the table in those first weeks before we diagnosed the drop-off. [Edit: your own reflection.]

I didn’t make the ask smaller. I made it worth saying yes to.