All work
Case Study · 2021–2022 · Dushanbe City Bank

DC Next
a bank and a city
in one app

Replacing a legacy wallet with the bank's main digital product — payments, transfers, transport and 300+ city services, rebuilt around the two things people actually open a banking app for.

Role
Lead designer — team of two, ~90% of screens and decisions
Duration
2021–2022
Status
Shipped · still live, evolved since
Company
Dushanbe City Bank

The old app treated everything as equally important — so nothing was.

Dushanbe City Bank launched DC Wallet in 2018. By 2021 it was the bank's public face and its main channel to customers — and it was built like a folder of shortcuts. Eleven tiles of identical weight. No balance anywhere on the home screen. Paying for something sat next to "Favorites" as an equal.

The brief was to replace it with DC Next — an app that had to hold ordinary banking, city services, utility bills and public transport in one place, for a customer base that ranged from daily smartphone users to people opening a banking app for the first time in their lives.

I led the design end to end: information architecture, all core flows, the visual system and handoff. We were two designers; roughly 90% of the screens and the structural decisions were mine.

300+
services under one navigation model
50+
countries reachable for transfers
11 → 4
equal tiles became four primary actions
1 QR
primitive covering three different jobs

Two questions the old home screen couldn't answer.

People open a banking app for one of two reasons: to see how much money they have, or to pay for something. DC Wallet's home screen answered neither without a tap.

DC Wallet · 2018–2021
  • No balance on the home screen at all — it lived two taps deep.
  • Eleven tiles of equal size and weight; "Pay" ranked the same as "Favorites".
  • The bottom of the screen held a title and two icons — no persistent navigation.
  • Services opened as a flat alphabetical list, the same for everyone.
DC Next · shipped 2022
  • Balance and bonuses sit at the top of the home screen, above everything else.
  • Four primary actions instead of eleven equal tiles; the rest moved into sections and a searchable catalog.
  • A permanent five-tab bar with the QR scanner as its center action, reachable from any screen.
  • Favorites fill themselves from what you actually use, and stay editable by hand.
DC Wallet legacy home screen
DC Wallet — the legacy home screen: a grid of equals, no balance in sight.
DC Next home screen
DC Next — the balance is the largest element on the screen; four actions, favorites and recent recipients follow, with the scanner anchored in the center of the tab bar.

No usability lab — but there were four years of real usage data.

I want to be precise about this: there were no moderated user interviews and no usability testing sessions on this project. What we did have was better than a guess and worse than a study — the bank's own usage data from the app it had been running since 2018, plus the operational knowledge of the teams handling support and payments.

Frequency isn't evenly distributed

A small set of services carried the overwhelming majority of transactions — mobile top-ups, utilities, transfers. Those earned a place on the home screen. Everything else went into a catalog with search, instead of competing for the same visual space.

…but every customer has their own small set

Ranking by aggregate frequency fixes the average user and fails the specific one. Someone paying the same kindergarten bill monthly gets no benefit from a global top-nine. That contradiction is what Favorites exists to resolve.

A first-time user and a daily user share one screen

The customer base spanned people paying with QR every morning and people installing a banking app for the first time. The home screen had to be immediately legible to the second group without slowing down the first.

Four decisions that shaped the rest of the app.

01 · Balance above everything

The card balance, the currency and the bonus total moved to the top of the home screen as a single block. It sounds obvious written down; it was the largest structural change in the redesign, because it demoted the service grid from "the home screen" to "one part of the home screen" — and that demotion is what made it possible to cut eleven equal tiles down to four primary actions.

02 · One scanner, three different jobs

Paying a merchant, topping up a wallet by QR and paying a bus fare are three different mental tasks — but technically they are the same act: scan a unique code, confirm an amount. Rather than build three entry points, the scanner became the center button of the tab bar — the only round, colored element in the navigation, present on every screen — and the flow after the scan adapts to whatever the code turns out to be.

01
Scan
One button, always in the same place — no need to know in advance what kind of payment this is.
→
02
Recognize
The code identifies itself: merchant, wallet, transport or terminal.
→
03
Confirm
Amount, source card and recipient in one view — the same layout in all four cases.
→
04
Receipt
Saved to history, repeatable in one tap, addable to Favorites.
Honest attribution: the unified QR standard itself was the bank's payment infrastructure, not my design. What I owned was the decision to expose it as a single always-present action instead of three separate features, and the interface of the flow that follows the scan.

03 · Favorites that fill themselves

A section that starts empty stays empty — most people never curate anything. So Favorites populates automatically from what the customer actually repeats, and remains fully editable: remove what you don't want, pin what the algorithm missed. The global ranking serves the average customer; this serves the specific one. Two mechanisms, one screen, no contradiction.

04 · Statements, version two

The first statement screen shipped as a correct list of operations that nobody could read quickly. The second version separated the statement from the operation log, added grouping and filters, and made a single transaction expandable into full detail with a repeat action. Two versions of the same screen exist in the file — I'd rather show both than pretend the first one was right.

Six surfaces that carry the product.

Everything below shipped and is still in production as the foundation of the app — later generations of the product added new layers on top, but the navigation model and these flows are the ones I designed.

DC Next — home
Home — balance first, four primary actions, auto-filled favorites, scanner in the tab bar.
DC Next — transfers
Transfers — six named directions, each stating where the money lands and what it costs before you tap.
DC Next — payments
Payments — the catalog that absorbed 300+ services, search above everything else.
DC Next — service payment
Paying a utility bill — check the account first, then pay. The same shape for all 300+ providers.
DC Next — credits
Loans — a tranche opens over the loan list: amount due, date, pay and schedule without leaving the screen.
DC Next — my cards
My cards — the card itself is the object on screen, not a row in a settings list.
DC Next — transfer to a DC Next wallet
Transfer to a DC Next wallet — checking the recipient is a deliberate separate step before any money moves.
DC Next — transfer to a Sber card
Transfer to a Sber card by phone — identical layout, different rails, with the fee stated above the amount.
DC Next — payment history
History — grouped by day, every line carrying card, counterparty, amount and status.
DC Next — CityCard transport card
CityCard — the city transport card as a card: tariff, balance and ID, topped up from the same app that pays the bus fare.

"Lend to a friend" — the feature that made the app do the asking.

The transfer form plus one field: a due date. From there the app reminds the borrower a few days before and again on the day itself — so the lender never has to be the person who asks.

The three chips — three days, a week, a month — are there because that is how lending between people is actually agreed out loud. The hard part of lending to a friend was never the transfer; it was asking for it back. Moving that to the app is the whole idea. It was designed in full and never shipped: priorities moved elsewhere, and at the time that was a reasonable call by the bank. I show it because a portfolio of nothing but wins hides how the work actually goes.

Lend to a friend flow
Designed · not shipped

DC Next replaced DC Wallet and became the bank's main digital product.

REGISTERED USERS
~1.7M

registered users on an app serving a country of roughly ten million people.

MONTHLY ACTIVE
~810K

monthly active users completing three or more transactions — not just opening the app.

DOWNLOADS
1.46M

total downloads, running at 2,500–3,000 per day at the time of measurement.

About these figures: they come from the bank's own reporting and describe the state of the product in 2023 — a year after my involvement ended, so they reflect the work of the team that continued it as well as mine. They were shared with me directly and are not published on a public page I can link to. I'd rather flag that than present them as independently verifiable.

What came after me

The app kept growing after 2022 under the team that took over: a mini-apps layer with third-party services, face-unlock security, sign-up through Telegram, parking and EV-charging partners. None of that is my work. The navigation model it all sits on — balance first, permanent tab bar, scanner in the center — is.

What I'd keep, and what I'd do differently.

What worked

  • Ranking the home screen by measured usage, not opinion. It ended the recurring argument about whose department deserved a tile.
  • Treating the scanner as navigation, not a feature. Putting it in the tab bar meant every later payment type inherited an entry point for free.
  • Automatic Favorites. Personalisation that requires no setup is the only kind most people ever get.

What I'd do differently

  • Insist on talking to users. Usage data tells you what people did, never why they stopped. On a product this size, even ten interviews would have changed decisions.
  • Design the statement screen last, not first. I built it early from the data model and had to rebuild it once real transaction histories made it unreadable.
  • Record the reasoning as I went. Four years later the screens survive but the arguments behind them don't. I now write decisions down while they're still being made.

See the app that came out of it

DC Next is still live on both stores — several product generations past the version I designed, built on the same navigation model.

→ Google Play → App Store → All case studies