Purno

A first-time-friendly retail app that raised a $10k seed and cut transaction time ~20%

Purno is an app for small retail shops in Bangladesh. The owners handle several payment methods, tools that do not connect, and little tech experience. I was the first designer on it, and I took it from a loose scope to a funded product.

$10k seed

The MVP I designed raised a seed round

~20%

Faster transactions from the color coded cards, in moderated testing

Cross platform

One app consistent across devices

Localized

Made for first time app users in Bangladesh

Role: First product designer, working directly with the founder
Timeline: 2 months
Platforms: Desktop, tablet, and mobile

Three decisions that shaped the product

01
Flatten the category structure
The structure nested categories four levels deep, which slowed down a shop owner with a customer waiting.
02
Test five product cards at arm's length
The selected state had to read from across a counter. Moderated preference testing picked the version the home screen was built on.
03
Tie identity checks to payment activation
Verification takes effort, so it comes at the moment a merchant has a reason to finish it.

Comparative UX auditInformation architecturePaymentsDependency mappingModerated testingLocalized UXCross platform designDesign led testing

Behance

Problem

Payments and sales did not connect

Shop owners here take cash, bank transfers, mobile banking, and wallet payments. Each runs on its own. Purno pulls payments, inventory, sales, and reporting into one system.

The app had to work for owners new to technology. Ease of use, speed, clear menus, and language support came first.

Purno, the merchant app

Research

A comparative UX audit set the feature order

The founder started with an unfinished scope and a few screenshots. I ran a comparative UX auditof Square, Loyverse, Toast, Clover, Lightspeed, and bKash. I was looking for patterns people already knew, and complexity to avoid.

Feature benchmarking against each competitor shaped the design decisions.

The research split in two. The founder knew these shops and the people in them, so he was the subject matter expert and the one who brought shop owners in. I ran the sessions with them and pulled the findings together. Most were running a business app for the first time and were not confident with software. That set the standard for the interface.

What came out of that shaped the iterations. I wrote the onboarding wording, cut the menu depth, and limited what fits on one screen, so there is less to take in at once.

Comparative UX audit

User flows

Information architecture

Purno is mostly one app, the merchant app a shop owner runs all day. A payment app runs alongside it and stays out of this study. The black theme build appears at the end.

Process

Components before screens

Dependency mapping set the order. Inventory management and the path from home to cart came first. Then each payment method got its own flow.

Most screens started from a competitor. Design pattern analysis let me take the best version of each pattern and shape it for the shops here.

I designed components before screens, following an atomic design approach. They acted as variants, and feedback rounds picked which one to keep. Most screens took two or three rounds.

Core flow

Four screens carry a whole sale

A shop owner signs in, finds the product, takes the payment, and closes with a receipt.

Products come first. The home screen opens on color coded categories. The scanner sits in the middle of the tab bar for anything with a label.

Payment is a page of its own, opened from Checkout once the cart is final. Cash, Card, and MFS sit there as three entry points, with Split Amount on top.

The receipt carries the transaction ID, the time, and the method, then the order amount, discount, and fee.

Find the product

Take the payment

Close the sale

Payments

All payment methods in one place

Card and mobile wallets

Card runs through tap or insert, a 4 digit PIN, then processing. The tap screen and the PIN pad copy a real POS terminal. Shop staff have used a card machine before, so there was nothing new to learn.

MFS covers bKash, Nagad, Rocket, and Upay in one flow. The merchant picks a provider, generates a QR, and the customer scans it. Payments default to Bangla QR. The chosen provider stays the default, since shop owners keep one wallet. A new provider is one more line in the picker.

Card: tap or insert, PIN, processing

MFS: pick a provider, generate the QR, wait for the scan

Business verification

MFS stays Unlinked until the merchant finishes four steps: NID front and back, trade license, payment setup, and a PIN. Both documents use the same camera frame.

The trade license needs more. The system reads the owner name, company name, and license number off the photo. Typing a long registration number on a phone is where a first time user gives up.

I tied the identity check to payment activation deliberately. A merchant taking mobile payments is a verified business first.

The four steps, the document capture frame, and the settlement account

Transaction detail

The old screen led with a date. It carried Refund and Checkout buttons on a record that was already closed.

The record cut back to what a merchant asks for

Inventory

Finding stock and getting it into the cart

A shop owner works these screens with a customer waiting. Every one was redrawn around that.

A flat category list

The information architecture nested categories inside categories. Sewing materials held Pins and Strings. Each of those held two more levels. A category screen mixed subcategories and products together.

I flattened the structure to one level. Square nests categories, like most global retail apps. That depth suits staff who use one app all day. These shops sell a wide mix of items, so cutting the nesting was the trade worth making.

Three levels flattened to one

Quantity by item type

Products are counted two ways. A packaged item like Lay's asks for a pack size, then a stepper with +5, +10, and +15. Bulk items like blueberries take a weight in kilograms, with +5, +10, and +25. Bulk moves in bigger jumps, so the shortcuts match.

Pack size and a unit stepper for packaged goods, direct weight entry for bulk

Five versions of the product card

The product card settled before the home screen. I ran moderated preference testing on five versions, each in its default and selected state. The selected one had to read from an arm's length. The home screen was built on the version that won.

Five product cards, top row default, bottom row selected

The home screen

The first home screen used emoji chips and turned the whole card purple when an item was picked. Merchants already used color to group their categories, so I moved that role to the card.

Color coded cards cut transaction time by about 20% in moderated testing. The structure stayed as it was. A visual cue did the work.

Emoji chips replaced by color coded cards. The annotations carry the test note behind the 20%

The barcode scanner

The cart sat in a drawer that covered half the camera and stayed open. The scanner could not show what had been added.

The cart drawer moved behind a basket icon

Screens

The rest of the merchant app

Sales reporting runs on three time frames, so the picker changes shape for each. Daily opens a calendar. Weekly lists the weeks with a checkbox on each, so two can be added together. Monthly drops to a year.