← All work Case study · Live product

dayexam.com

An exam-prep platform for SSC, HSC, BCS and admission tests — built for a client with zero technical background who handed me the entire decision space.

RoleProduct & UX design, solo
BuiltWeb → admin dashboard → mobile app
StatusLive · 3.5K+ students enrolled
dayexam.com exam listing page
01

The brief

"I only know teaching. You know website, app, and dashboard design — you decide what's right."

The client was a teacher, not a technical person. He wanted one thing: a place where students and guardians could log in, buy courses or exams, and take tests. Beyond that, he left research, decisions and testing entirely to me — his only role was to approve or reject at the end.

02

Directing AI, and what I built

I did the market research myself — Bangladeshi student and guardian behaviour, what they trust, what puts them off — then fed it to AI to cross-check and fill gaps, judging the output as a Bangladeshi user, not just a designer. AI's first flow ideas were wrong: I rewrote the core flow by hand seven times before it held up, then stopped asking AI for flows altogether and started drawing them myself, handing AI a finished flow plus a token system to build from.

Role-flexible accounts

One login moves between guardian and student — a parent can buy a course for their child and separately sit their own exam (BCS, bank job), without two accounts.

bKash + card, both

Fully online business, no in-person cash option — bKash covers the majority of student users, card covers the rest.

Split grading

MCQs auto-score with wrong-answer deductions; written sections are graded manually by the assigned teacher — no AI grading, since the client's budget didn't cover it.

Web → dashboard → app

Shipped the website first, then a client-requested admin dashboard, then a dedicated mobile app once the client saw the responsive web version working — one design system across all three.

dayexam.com home page

The home page — the entry point for both guardians and students.

03

The real problem

"Let people buy and take exams" was hiding three unresolved shapes. First, the account model: a guardian and a student aren't two user types, they're roles one person moves between, sometimes for different exams at different levels. Second, AI's payment flow stopped at "select course, pay" — no post-purchase view of "my exams," no result history, no answer for whether a retake is free. I raised the retake question directly with the client and got the real business rule (already paid once, so free retake) instead of guessing. Third, the grading split had to route written answers to the same teacher who set the question, given the budget.

04

What I rejected

I rejected AI's first flow outright and rebuilt it by hand across seven revisions — it kept treating "buy a course" as the end of the flow instead of the start of the exam-taking journey. I also rejected letting AI touch grading, even after the client himself asked whether AI could summarize written answers: no budget for the integration, and a wrong auto-grade on a real exam is worse than a slower manual one.

05

Outcome

Live at dayexam.com, both web and app. The client was satisfied enough to greenlight the dashboard, then the app, off the back of it.

3.5K+students enrolled
3surfaces shipped — web, dashboard, app
7hand-rewritten flow revisions
What I got wrong

My first flow draft ended at "purchase complete" — I hadn't designed what happens immediately after someone buys, i.e. where they land and how they get into the exam. The dev team caught it when they tried to build it and asked me directly. I fixed it by documenting the flow properly first, then handing over a visual design with tokens and a guideline, working through several rounds of meetings with an external dev team I didn't manage.

Next case study Terravue →