Product Design
A concept redesign of the monthly cashback picker in T-Bank, a Russian digital bank. Every month customers choose 4 bonus categories, but the screen only shows percentages. I made it show what each category is worth in rubles, without changing a single rate or rule.
The problem in one line
What the customer sees
“25% Shopping in Gorod”, “24% Tickets in Gorod”, “5% Cinema”. The percent comes first, in the same weight as the name.
What they have to work out
Which categories only count inside Gorod, how much they really spend in each, and whether 1% on all purchases deserves a slot.
Why it matters
The choice locks for a month. Pick by the loudest number and the card pays almost nothing, and people blame the card, not the screen.
It’s real and checkable
Rebuilt from the live T-Bank app, November 2026: the same 8 categories and rates. Screens are translated from Russian.
It’s about money, every month
Cashback is one of the main reasons to keep a card as your main one. A pick that pays little makes the whole card feel worse.
Customers complain about it
Reviews mention the monthly chore, categories nobody uses, and cashback that never arrived because a shop’s code didn’t match.
Insight 01
Percent comes first. The eye picks 25% over 1%, whatever you spend.
Insight 02
Half the list only pays inside Gorod. The condition hides behind a “?”.
Insight 03
The bank knows your spending. The screen doesn’t use it.
Insight 04
“Done” locks the whole month in one tap, with no total and no limit.
It worked if
Cashback earned per customer goes up, more people pick before their first purchase of the month, and “cashback didn’t arrive” chats drop per 1,000 customers.
It’s wrong if
Cashback goes up but card spending and retention don’t. Then customers simply cost the bank more, and the forecast needs a different goal.
01 · The pick screen
Before · the screen today
1. Percent comes first, so 15%, 24% and 25% win the eye.
2. Gorod-only categories look like the rest. The catch is behind “?”.
3. Where she really spends looks weakest: 5%.
4. Done locks the month. No total, no limit.
After · step 1
1. A ruble forecast leads each row. The percent moves to the subtitle.
2. Sorted by value, with one-tap “Pick by spending”.
3. Gorod-only categories say so and show 0 ₽.
4. Total, limit and an honest tip: fuel through Gorod pays most.
Why
The bank already knows what each customer spends. Showing it at the moment of choice turns a guess into a decision. Same 8 categories, same rates, same rules.
02 · What happens next
Step 2 · Category terms
Opens from “?” on a category: a one-line rule, the customer’s own spending and what paying through Gorod would earn.
Step 3 · Confirm
New. Opens on “Done”: picks, rubles, total and limit before the month locks.
Step 4 · Month progress
Later, from the Cashback section: earned vs forecast, and why a purchase didn’t count.
Why
Each step answers the customer’s next question: what’s the catch, what am I locking in, and is it working? The Gorod tip sells the bank’s ecosystem honestly instead of hiding the condition.
≈ 60 ₽ a month
Picking the four biggest percentages. Three of them pay nothing unless you buy through Gorod.
≈ 560 ₽ a month
Picking by real spending. The 1% on all purchases does most of the work.
≈ 1,365 ₽ a month
The same picks plus paying for fuel through Gorod. Better for the customer and for the bank.
1) A/B test
Black card holders, 50/50, for two monthly cycles. Primary metric: cashback earned per customer and picks made before the first purchase of the month.
2) Guardrails
Card spending and retention after three months, Gorod usage, and support chats about missing cashback per 1,000 customers.
3) What would change my mind
If cashback goes up but spending and retention don’t, the bank just pays more. Then the forecast should nudge toward spending goals, not only show numbers.
