Case study - Brief
“Silence has texture. Strength doesn’t need to speak. True design emerges when form and material align, when every surface serves purpose. Matte becomes the voice of restraint, shadow becomes structure. In the language of black, there’s no excess, only the weight of intention.”
Building for Bharat
The art direction focused on a monochromatic language defined by black on black contrast and elemental composition. Every surface was designed to feel tactile, from volcanic textures to the soft sheen of brushed glass. The aim was to capture stillness and presence, a brand that exists in silence, yet dominates through form and weight.
Photography and 3D visualization were approached like sculpture, treating light as material. Shadows reveal as much as they hide, defining contours with restraint. The visual tone echoes high end minimalism found in architectural spaces and modern craft, honest, bold, and physical.
Before we talk about the process, here's product at a glance.
The Gap
In 2017, India introduced GST — one of the biggest tax reforms in the country's history. For large businesses with accounting teams, this was manageable. For India's 63 million small businesses, it created a problem most small businesses weren't equipped to solve.
Tally — was built for accountants, not owners. To get anything useful out of it you needed to hire a dedicated person or take a course. Want your balance sheet? Ask your CA. It ran only on desktop — locking out every business without a computer.
Vyapar — had the right instinct but poor execution. Weak UX, high pricing, no cloud sync, desktop fully paywalled, mobile an afterthought.
Paper billing — the reality for the majority. Fast and familiar. But invisible to GST, impossible to reconcile at scale, and completely manual.
What we set to do
Make digital billing as easy as sending a WhatsApp message.
We built our design foundation on four core principles:
— visually clean but frictionless UX,
— zero cognitive load (no accounting knowledge required),
— an accessible entry point, and rapid time-to-value.
Creating a bill had to take fewer steps than writing one by hand, or paper would win
every time
But…
Adapting non tech people to use an app for a task they had done by hand for years… wasn't that easy as we thought…
Who We Were Building For
Our user base was incredibly diverse. On one end of the spectrum, we had 20-year-old tech-savvy. On the other end, we had 60-year-old shop owners
From our interviews, we grouped our users into three main personas:
The Power User (P3)
Highly tech-savvy. They explore new software easily, read the interface quickly, and want to move fast.
The Guided User (P2)
Comfortable with basic apps, but easily overwhelmed by cluttered screens, small buttons. They need the interface to be obvious.
The First-Timer (P1)
Zero exposure to business software. They rely entirely on physical ledgers and local vocabulary, and need the absolute lowest barrier to entry.
And, since most of our users were coming straight from pen and paper, we dropped all the accounting jargon and just used the words they use in their shops every day.
This created a massive design challenge: How do you build an interface that holds the hand of a beginner without frustrating a power user?
If we made the app too complex, our first-timers (P3) would instantly abandon it. But if we over-simplified it—forcing users through annoying, unskippable tutorials or excessive hand-holding—our power users (P1) would feel slowed down by the obvious friction.
Our solution was progressive disclosure. We kept the surface layer simple. The primary actions focused on what user needed the most, Make an Invoice. However, we kept the advanced, neatly tucked away just a tap or two deeper.
I made the platform incredibly easy to start using, but gave it enough depth to scale alongside the user as they became more confident.
Version 1
Clean, intuitive, modern, and built with the best intentions
I designed an interface that I truly believed was effortless to use.
WHAT WE DISCOVERED
The design was working. Just not for our actual users.
Few months in, something wasn't right. Support tickets kept coming with the same themes. I was watching SmartLook session recordings — sometimes hours at a stretch — and I kept seeing users hesitate, click on the wrong things, miss features entirely. So, decided to call users directly to address what was wrong.
The user calls gave me pieces that didn't make sense -
— All the information is cluttered, scattered and hard to read
— There are no lines to differentiate between sections
— I don't know where to click
— Lots of missing functionalities and features
— Features are incomplete
But I couldn't piece it together from recordings and calls alone. So I decided to go see for myself.
We visited actual users in Bangalore. I sat across from a shop owner and watched him open myBillBook on his actual monitor — not mine, not a modern display, not anything close to what I designed on
The monitor was old. The colour accuracy was poor. Everything I'd designed with subtle, light-coloured borders to keep things clean? Those borders were invisible. The dividers I'd used to separate sections didn't exist on his screen. The input fields didn't look like input fields.
What he saw wasn't a clean, modern product. He saw text clumped on top of text, with no visual structure holding it together. No dividers. No boxes. Just information with nowhere to go.
I went back to the office and pulled the actual resolution data with the PM.
75% of our users were on a resolution of 1366×768. Not the 1440×1024 most common for SaaS webapps. The content I'd assumed was visible above the fold was hidden for the vast majority of our users. The lighter colours I'd used for premium look were completely invisible on budget monitors with poor to no colour accuracy.
I had designed for a screen that most of our users didn't have!
THE rework — EVERY DECISION GROUNDED IN WHAT I'D LEARNED
I started over from first. Not what looked good on my screen, but what actually worked on theirs
What I found:
Borders invisible on low-accuracy monitors. Secondary CTAs, input fields and bordered sections indistinguishable from static text. Critical content cut off below the 768px fold.
What I did:
Increased all border weights across the product. Rebuilt the colour system with higher contrast ratios throughout, nothing that depended on accurate colour rendering. Redesigned every CTAs, input fields, sections to have unmistakable visual weight. Moved to a darker, more tone — it fixed the visibility problem and made the product felt it had more weight, which mattered because we were asking people to trust us with their financial records.
The trade-off I made:
Heavier visual treatment risked feeling less refined on modern screens. I tested multiple weights to find the threshold — visible on old hardware without feeling clunky on high-res displays.
Impact at glance
The rework didn't add anything new. It made the existing product usable.
Before the redesign, the majority of inbound support tickets shared the same root cause — users couldn't find things, couldn't read things, or couldn't figure out what to do next. Within a month of rolling out the new visual system, those tickets dropped by nearly 70%. Not because users suddenly became more patient. Because the product stopped fighting them.
~70%
Drop in UI-related support tickets within the first month of rollout
2024
31% → 64%
Users creating their first invoice within 24 hours of signing up
2024
+38%
More users discovering 3+ features in their first week
2024
+24%
Week 1 retention improvement after redesign rollout
2024
The deeper signals came from the data.
First invoice created within 24hrs of signup went from 31% to 64%. That number mattered more than almost anything else — because a user who creates their first invoice on day one retains at a fundamentally higher rate than one who doesn't.
Feature discovery was the one that surprised me most. Before the redesign, users were largely sticking to the features they found first and never exploring further. After restructuring the navigation around task mental models and adding text labels, the percentage of users discovering at least three distinct features in their first week increased by 38%. Features that had been sitting unused — inventory, reports, party ledger — were suddenly being found.
Retention improved by 24%. Before the update people complained the product had no basic features,
I'm not claiming the redesign alone caused all of that. Plenty of things went right at once. But the product stopped being its own obstacle — and that mattered more than anything else.
