RoughSheet

Growth with no paid marketing

Campus ambassadors

Two per large college and one per smaller college, across 30+ NITs from year 2 and the 40 Pune colleges, compensated with an internship certificate and career guidance from the co-founders.

Placement officers

In 2016 to 2017, all 40 targeted Pune colleges agreed to recommend RoughSheet officially. The product was free, met a real student need and carried no risk for the college.

Café partnerships

With no budget for rewards, the team mapped cafés around the 40 Pune colleges, assessed footfall through Zomato and Google ratings, and approached the third- and fourth-rated cafés, which were good and wanted more student traffic. Cafés gave 10% to 25% discount coupons at no cost, and the weekly top three practice-sheet performers received them.

Referrals

Every user had a referral code and ~33% used theirs. Among referring users, referral productivity rose from 1.1 to 1.5, mostly through invite links shared in WhatsApp groups, giving a blended viral coefficient of ~0.43 against 0.1 to 0.3 for typical consumer products. Placed students kept recommending the product to juniors, which created a referral loop between cohorts.

Features from interviews, analytics and experiments

Daily practice sheets

Timed, 10 questions in 15 minutes. They began as a mixed set and became one sheet per subject: 60 minutes and 40 questions a day in total.

Rank list, from user interviews

Weekly and per subject, based on that week’s results, with at least 2 results per subject needed to qualify.

Study material, from user feedback

Launched in year 2, by subject and chapter: concepts with text and diagrams, then three untimed exercises of 25 questions each at easy, medium and hard levels. It became the most-used feature by time spent.

Mobile responsiveness, from analytics

In 2014 to 2015 many mobile users started a sheet, answered nothing, and tried another subject, repeatedly. Nobody raised it in interviews. The site rendered badly on mobile browsers and native apps were beyond the team’s resources, so we rebuilt the site to be responsive. Practice-sheet completion on mobile rose from ~20% to ~80%.

Progress index, from behavioural data

Users repeatedly opened chapters and exercises to find where they had stopped, despite a “resume your studies” link on the home page, which inflated engagement figures. An index with tick marks for completed work cut session time from over 20 minutes to ~18, and follow-up interviews confirmed the improvement.

Weak area analysis, from user interviews

A clear breakdown of strengths and weaknesses, with a notification when it changed.

Planner, from interviews and research

Based on my reading of psychology and cognition research on study planning. Adoption (planning at least twice a week) rose from ~75% to ~85% through iterative UX changes.

Result sharing, from experimentation

A shareable result poster after each sheet, with comparison against friends. 75% of users who completed a sheet shared their result. It brought in new users, and its effect declined over time.

Calling users every week

I spent about an hour every weekday and two hours each weekend day calling users. On weekdays I called 4 to 5 users a day (~25 a week) who had been active for 30+ days and then silent for 7+ days, for 15+ minutes each. At weekends I called ~15 users who had tried the product and not returned after 7+ days. That came to ~8,000 calls over ~4 years. In the early years every churned active user was reached; later the calls focused on the most recently silent users.

The calls were for feedback. At least 15% of early drop-offs returned after a call as a by-product, and users who were called were more likely to refer others.

Measuring an active user

An active user met any one of three conditions: more than 3 daily practice sheets a week (of 28 possible), more than 30 minutes a week on practice sheets, or more than two logins a week. Internal data showed that users meeting any of these returned consistently. Active users completed 1.2 sheets per session across 3.6 sessions a week. Over 80% of usage was on weekdays, much of it during lectures, with phones held under desks.

Users typically prepared for ~45 days and stopped once placed, a lifecycle identified in data and confirmed by users. New annual cohorts, varied placement calendars and the daily practice habit kept usage steady across the year.

The B2B move and shutdown

In the fifth year I approached 35+ companies in Pune with a recruiter-facing offering. Smaller companies wanted narrowly defined candidates, which suited the product’s data; larger companies were bound by long contracts with established assessment providers. The team could not close deals in time. The late move into B2B, together with team fatigue, led to the shutdown in June 2018.