A loyalty offer for the customers most likely to leave

What I designed

Rather than showing one screen in isolation, I walked people through the full program as it would unfold: the month they qualify, the month the reward applies, and the month after. A loyalty program is a promise made across time, and testing a single screen cannot tell you whether the promise holds.

Recruiting was the hard part, and worth saying plainly. The internal customer list returned nothing from a hundred attempts. I built the sample across three separate sources and screened on send corridor and monthly send volume.

What I found

The reward was never why anyone opened the app. Attention belonged to the transfer in progress. The offer got scanned quickly for anything that looked like a security problem, then dismissed so people could finish what they came to do.

The structural problem was larger. Nobody could tell how long the program lasted or what the rate would actually be, and people cannot plan sends around a promise they cannot size. For an audience that budgets carefully and sends on a schedule, that turns a loyalty program into a trust problem.

It feels overwhelming to think about how much money I'm sending home. Where did my money go?Participant, US to Philippines

What it changed

Design changes to placement, clickability, and progress feedback, plus the value proposition the team had buried in the copy: the better rate applied to every transfer, not one. People who noticed it were surprised and pleased, and most never noticed it.

The findings fed straight into the A/B test design. The study also surfaced a set of high-amount sender questions the team had not scoped, including heavy desktop use in a segment the app had not been built for.