Week one

I joined Masterschool on a Monday in December 2021. By Friday I was designing the marketing site that schools would launch on.

There was no one to ask permission. There was no design system. There was a deadline — the first cohort was three weeks out.

I learned what founding PM meant the way you learn it: by doing the next thing that wasn’t getting done.

The arc

Bootstrapped, sub-100, pre-funding when I joined. By the time I left in July 2024, the company had closed a $100M seed and operated in 20+ countries. Through that arc I owned, in rough order:

The story

Positioning before roadmap. A one-pager that named who Masterschool was for, why a network of practitioner-led schools beat a single bootcamp, what we believed about career mobility. That document framed the seed deck, the website copy, and the first product hires. Some weeks, the most useful thing I shipped was a paragraph rewrite.

The platform

A white-label site that schools launched on. Track landing pages. The framing for the Master ("you don’t teach a course, you launch a school"). Community surfaces that respected the cultural distinctness of each school.

None of this was a roadmap. It was a queue of what’s on fire this week. The boundary between product and marketing was fictional. Between product and design, also fictional.

Why we shipped a Chrome extension at all

You don’t see ed-tech ship Chrome extensions much. The reason it was the right move at Masterschool says something about the shape of the problem.

A learner in our program was a working adult, often with kids, often with a day job, transitioning to a new career. They studied in fragmented blocks: 30 minutes between meetings, an hour after the kids were asleep, a Saturday morning before chores.

The platform we shipped at first assumed continuous attention. Open the LMS, watch the lecture, complete the exercise, move on. Reality didn’t look like that. Students opened the LMS, watched two minutes of a lecture, got pulled away, came back three days later having forgotten where they were.

The Chrome extension was the answer to meet the student where they actually study.

It pulled the assignment that mattered most that week into a small panel that opened with a keyboard shortcut. It surfaced the next mentor session as a notification that wasn’t trapped behind a login. It tracked when the student was reading documentation related to their curriculum — and quietly counted that as practice toward their streak, even if they never opened the LMS.

We shipped it in twelve weeks. By month four it was the most-used surface on the platform. Not because it was a better LMS. Because it acknowledged that the student didn’t live in our LMS. They lived in their browser, their day, their fragments.

The surface that wins is the one that meets the user where they already are. Not where you want them to be.

The retention problem nobody could see

When I started on retention, I read everything. The bootcamp churn literature. The Coursera completion-rate papers. The Lambda School post-mortems. Most of it pointed at content difficulty, time commitment, financial stress.

We had data on all of those. None of them explained our churn well.

What did explain it, eventually, was a sentence a student wrote in a community thread two months in:

I keep waiting for someone to realize they made a mistake admitting me.

I read 200 more posts that week. The same sentence kept showing up, in different words. Everyone here is smarter than me. I’m the one who’s not getting it. I should have figured this out by now.

The students dropping out weren’t dropping out because the work was too hard. They were dropping out because they didn’t feel they belonged. The work being hard was just the proof, in their head, that they shouldn’t be there.

This is impostor syndrome at scale. And it’s particularly cruel for career-changers — the population our entire platform was built to serve.

Three mechanics that worked, together

We didn’t ship one feature. We shipped a system.

Streaks, but not the kind you’ve seen on Duolingo. Not "30 days in a row." Streaks tied to placement-relevant behavior — a streak for completing your weekly mentor session, a streak for submitting your interview-prep assignment on time, a streak for showing up to the cohort meeting. The unit of progress was something a future employer would care about. And critically, the streaks were visible to your cohort. Not as a shaming leaderboard. As a low-key signal that everyone was showing up.

A cohort ladder, with one design call: nobody saw the bottom of the ladder. You saw your position relative to the median. You saw who was just above you and just below you. You did not see "you are #47 of 50." Because the goal wasn’t to motivate the top performers. The goal was to show the imposter — the one staring at her screen at midnight convinced she was the only one struggling — that she was, statistically, in the middle of the pack. That her struggle was the normal one.

Milestone badges, but tied to outcomes, not effort. First mock interview. First real interview. First offer signed. First paycheck. Not "completed module 3" — that’s vanity. First paycheck is the thing the student herself, in three months, would describe as the moment her life changed.

We shipped these together. The streaks created visible belonging. The ladder reframed struggle as normal. The badges made progress legible to the student herself.

The number

Cohorts on the new system retained 15-20 percentage points better than cohorts that came before. We measured this carefully because, with ISA models, retention isn’t a vanity metric — it’s the unit economics.

The number wasn’t what made me proud of the work, though.

What made me proud was the threads, two months later, where students wrote things like I almost dropped out in week 4. The fact that my streak was still alive and that Yael’s was alive too kept me showing up. By week 8 I’d forgotten I ever wanted to quit.

We hadn’t built a motivation system. We’d built a belonging system. The retention was a side effect.

The closing line I keep coming back to

The deepest thing this seat broke me out of, looking back, was the assumption that retention is about content. It’s about whether someone, alone at midnight in week four of an Income Share Agreement, can look around and see they’re not the only one.

The title was Founding PM. The job was everything that breaks first. This one broke first because nobody was reading the threads.