CaseNexus.com was a P2P job interview platform for management consulting ‘case’ interviews. It removed scheduling, geographic and fear barriers to practice, to help them secure jobs. In this piece I demonstrate how I went from a design that required a lot of engagement, to one needing almost none (Active -> Passive), as well as a few design lessons I learned the hard way.
Management consulting is amongst the most popular career tracks for university students. Firms like McKinsey, BCG & Bain. Roughly 100K students apply globally each year. These firms have a unique ‘case’ interview format. Candidates are asked to solve hypothetical business problems like ‘why did a desk fan manufacturer’s profits decline 5% last year’. You can prepare for these interviews. Students are very dedicated, and start reading cases months before. Experts say however that live practise is key (and recommend doing as many as 50). I built a site to help students meet and practise online.
Part 1: Build, Build, Build
Without any process — no ‘ research, prototype, test’ — so began a quite monumental building spree. Initially I envisaged a website with a map — students would plot their location, see others nearby, and arrange to practise at a local coffee shop.
I must have done 50 different versions. I built an interactive list of users on the left, where you could browse and filter users — it was a crazy-complicated AJAX-enabled engineering marvel. It was so slick.
I called it ‘CaseNexus.com’ — ‘nexus’ means ‘hub’ in Latin. I spent way, way too long working on the logo. The pyramid was a nod to the McKinsey book ‘The Pyramid Principle.’
Once I had the map page looking good (V30), I felt strongly that it was not sufficient on its own to launch. Then I started ‘feature-creaping’ to total oblivion. In my mind the MVP required and one-stop-solution. First I built of library of practise cases, fully AJAX enabled, sortable — it was beautiful:
Then a full calendar system for each user — they could send case invitations to friends, and get reminders:
Next, a very sophisticated feedback system with a form to enter scores and graphical analysis of your scores. There were 3 different types of charts! Interactive and all.
After several months, now with ~25 features, I felt I had MVP and it was time to Launch. I arranged for it to be mentioned on the Cambridge Careers Dept mailing list — that day I got 1000 visitors, and 300 signups. What happened was something I will never forget. Of the 300, about 10 sent friend requests on the map, but not one did a case.
Part 2: Start Again
So I’d just wasted months building a useless monstrosity. I started again. Let’s just make one feature work. What was the top of the funnel? Meeting. And why am arranging for in-person — I bet these practise cases will work over Skype too! I’d seen on another site an ‘availability marketplace’ — users post when they’re free in the next 24 hours, and another user can book it. I built it, launched it, and had about 2–3 students doing 90-minute video calls every day.
Once paired up, 10 minutes before students were emailed a link to a WebRTC video call — it was awesome.
Part 3: Uncertainty
I let it run for a few weeks, and got some great feedback:
“Thanks to Casenexus I just got an offer from McKinsey!”
I surveyed my users via email — the majority specifically didn’t like the uncertainty of waiting for another user to book their time.
It seemed that what users really wanted was to be “guaranteed” a case partner at any time. This might be possible at massive scale, but not to start. Then I thought why don’t we just open for 1–2 times/day! I wireframed it out:
I re-built it. Users could opt in to a slot, and I would match them up manually [a ‘Wizard of Oz’ mechanism]. I re-launched. Right away a new problem emerged. 50% of users didn’t show up. Whereas before they felt beholden to another person [in a ‘marketplace’], there was no longer any real pressure to.
Part 4: So Simple
This no-show problem led to my favorite eureka moment of the project. I’d just seen the ‘Product Hunt Live’ landing pages, which have a count-down to a celebrity interview. Why don’t we just have one time each day, and to join users simply arrive a few minutes before? No booking in advance. Such a simple solution. I re-wire-framed:
Striving for passiveness, I then realized I could bring the entire product down to a one-click experience, using arguably the most powerful channel: email. I set up the site to email all users 10 minutes before the time each day, saying ‘would you like a case partner today?’, with a big green button to click if they did.
Clicking the button a few minutes before scheduled time took them to the landing page in their browser. From there they were automatically opted-in for case.
I re-launched, and users seemed to take to it without question. The no-show problem was fixed. Surveying users again however, another problem emerged. Some users weren’t using the site because they were afraid of being matched with someone much better than them. The switch from the marketplace model had taken away the user’s choice of partner. I concluded I could easily fix it with a simple skill selection menu on the landing page.
End-game & Retrospective
I’m sorry to say that at this point I decided to park the project. I could see that growing it was going to take passion, and having taken so long in the building, mine had died.
But I learned so much:
- Just make one feature and then test it with real users. If you’re thinking ‘I need 5 features for MVP’, then revaluate.
- Don’t code first, wireframe. I lost so much time solving engineering problems for features I didn’t end up shipping.
- Don’t hesitate to restrict options for people. Optionality is tedious for users — often they just want to be told what to do.
[You can return to the index of my portfolio here]