Most people did not try again after a failed payment.
Out of every 100 failed payments, 30 were tried again and 10 went through. And 2.5 per cent of users with a failed payment blocked their Paytm account.
Nishant Dubey
Senior AVP, product design. Design innovation in fintech.
I am at Wells Fargo, designing B2B fintech products and working on AI product innovation. I use AI tools through the whole of a project, from research to handoff. Behind that are nine years of payment flows and gamified journeys that moved the numbers.
What I do
GrowthFintechPlay
Across payments, games and banking, every project starts with the job a person is trying to get done and ends with a number that shows whether the design helped.
The index
Selected
Three projects, each written up in full.
Scroll to move along the shelf.
30% to 49%Tried again
04FlagshipProblem Insight Outcome
Confusion
People were skipping the reason for the failure.
A three-week heatmap study across more than 20,000 failed payments showed taps going to the logo, the amount and the floating bar, and very few on the reason.
The reason became the title of a bottom sheet, with the next step under it.
In December 2023, 49 of every 100 failed payments were tried again and 22 went through, up from 30 and 10.
Evidence
64Rise in the retry rate after a failed payment, per cent.
35Reduction in cost per acquisition, per cent.
20First time deposits, up from six per cent.
50New players from referrals, up from five per cent.
Teams I have built with
Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.
Antoine de Saint-Exupéry
How I work
I use AI at every stage of a project to move faster. Each stage still ends with me checking the work against real users, the design system and an agreed number.
01Understand
Start from the job people need done
Before I draw a screen, I write down the job people hire the product for and check it against interviews, tickets and analytics.
- AI speeds up
Sorting hundreds of interview notes and tickets into themes.
- Stays with me
Meeting users myself and reading the quotes behind each theme.
02Explore
Go wide, then cut to one
For each key screen I explore about ten directions, narrow them to three to work up in detail, and take the strongest one forward.
- AI speeds up
Drafting flows and layouts, so ten directions take a day.
- Stays with me
Cutting what is generic or does not serve the job.
03Test
Test early with real users
I test working prototypes with real data and every state, including the moment an AI answer is wrong, before engineering builds.
- AI speeds up
Turning a design into a working prototype in a few hours.
- Stays with me
Running the sessions and deciding what the findings mean.
04Measure
Agree the number before the design
With product and engineering, I set the goal, the signal that shows progress and the metric to read. Then the flow ships as a test.
- AI speeds up
Pulling results and session recordings into one summary.
- Stays with me
Reading the result, even when it says the idea failed.
What mentees say
I have given 65 mentoring sessions on ADPList. These are my mentees' own words, unedited. The sessions I run are on the mentoring page.
His in-depth feedback on my designs and his willingness to answer my questions helped me to see things in new and different ways.
Drag the top card, or use the arrow keys.
Review 1 of 10.
Growth designGamificationFintechDesign systemsUX researchHeatmapsA/B testingFigmaLottieRiveDesign opsMentoringGrowth designGamificationFintechDesign systemsUX researchHeatmapsA/B testingFigmaLottieRiveDesign opsMentoring