Leaf’s Winter 2026 Cohort Analysis
How you can help!
Reposted from EA Forum
Executive Summary
We ran five-week online courses for 368 teenagers selected from 3,901 applicants across 30+ countries. 87% earned certificates, investing 25+ hours each to complete the full course (our highest ever!), and for the first time, parent tuition exceeded cohort expenses, even while offering fully free-if-needed access, which ~40% of participants took advantage of!
Things are going well, but now is a critical time to scale our success! YOU can help if you are/ can recommend:
Organisations, competitions, schools, or individuals who can share our upcoming programmes with exceptional students interested in impact - especially in the US.
Collaborators who work with university/early-career students, towards whom we can direct the incredible talent we’ve been identifying and cultivating.
Connections with possible funders to support our work and the increasingly interconnected work of our partner program, Non-Trivial.
Here’s our one-page summary and complete data document on the recent cohort, including deeper context and more data.
What is Leaf?
Leaf works with exceptional teenagers who want to apply their talents to pressing global challenges. We run four five-week online courses (The Mathematics of Morality, Biology for a Better Tomorrow, Dilemmas and Dangers in AI, and History to Shape History), combining self-paced content, weekly discussion calls, problem sheets, projects, and speaker events. We select for talent, agency, altruism, and honesty/integrity using application questions about impact and puzzle tests normed against the general adult population; accepted Finalists average the 91st percentile and a meaningful and increasing proportion land above the 99th percentile.
Headline changes we’ve made from previous years:
Leaf and Non-Trivial are increasingly collaborating and sharing resources in the best interest of both organisations, given our shared focus on pre-university impact and complementary theories of change.
I have stepped into the Head of Courses leadership role, managing staff, participant-facing communications, feedback, and marking operational workflows, and helping Jonah with wider strategy for Leaf and Non-Trivial.
We pivoted from a donation model to a tiered tuition model with free-access-if-needed; We doubled our revenue while also increasing the percent of students receiving fully free access by 15 pp.
We opened up to our first fully international cohort after previously only allowing UK-based students to join previous Leaf Cohorts for logistical and legal reasons. This led to 90+ countries represented in applications and 30+ in the accepted finalist cohort.
Impactful results
Below are some results from outcomes we focus on most when assessing Leaf’s impact:
Reaching talented students early. Half or our Finalists report no exposure over the prior three months to 80,000 Hours, Astral Codex Ten, Non-Trivial, or Effective Altruism (EA; e.g., EA Forum, EA Course) and the share who report having spent more than 10+ hours with impact/rationality sources like these is only 6.4%.
Ecosystem onramping and referrals. Through our application process, we generated roughly 7,000 newsletter subscriptions for six partner organisations (80,000 Hours, Clearer Thinking, GWWC, Center for AI Safety, High-Impact Medicine, Rational Animations), estimated at $32-75K in value to those orgs. We also referred 43 of our graduates into Non-Trivial’s spring Research Foundations Programme, 20 of whom were “counterfactual finds” who hadn’t applied to Non-Trivial independently, and 23 who’d initially been rejected by Non-Trivial but got accepted after strong performance during Leaf.
Career influence grew. Reported “Career plans changed” rose from 4.7 to 5.1 on a 1-10 scale, with nearly half of Finalists reporting above 5. Lots of students described discovering career paths in AI safety, biosecurity, policy, and animal welfare that they just hadn’t known existed. 82.5% reported their views on the most pressing global problems had shifted. I think part of this is that students in the W26 cohort arrived more open-minded than S25 (fewer students reported being locked into one degree path), which probably made them more receptive.
International expansion. This was our first cohort open beyond the UK. We drew applications from 90+ countries and accepted Finalists from 30+ countries, which is exciting but also brought real friction. Actual enrollment upon acceptance dipped from 92% to 84% since timezone barriers hurt call attendance for some courses. Flexible scheduling is a clear priority going forward. This is still promising as a first attempt at international openness since many of our excellent international alumni can now go and spread Leaf to other excellent students in their communities.
Furthermore, these students go on to actually follow-through with pursuing impactful work! We continue to carefully track our maturing alumni base. See examples of recent alumni outcomes tracking here.
The two numbers I’m proudest of:
87% certificate completion. Last cohort, this was 68%. Every course improved by at least 10 percentage points, which really suggests something structural changed rather than one course having a good run. The “middle dropout” zone (students who submit 1 to 3 of 5 activities, then fade out) collapsed from 25% to 8%. This is also a good signal about Leaf’s ability to scale.
In large part, we attribute this to clear participant-facing communications from the facilitation team and me, as well as operational work I’ve done to cut out inefficiencies in our processes and user/facilitator experience regarding feedback and weekly submissions.
Tuition exceeded Costs. We had 320 certificate earners at effectively zero net cost per completer. We think the biggest driver was our new tiered tuition model (Free / Subsidised / Standard / Supporter). Importantly, 42% of families accessed the Free tier, so this wasn’t at the expense of accessibility.
The surplus helps cover offseason work like curriculum development, facilitator training, and legal/insurance costs that aren’t in the per-cohort estimates, but the trajectory is a genuinely strong signal for financial sustainability.
What’s next?
We want to scale the impact of future cohorts without compromising on our selectivity bar. It seems like Leaf is working well and creating impact in a hugely cost-effective way. The systems I worked on improving this time around are scale-agnostic, removing any obvious blockers to scaling our impact other than reaching more top students who meet our high admissions bar.
To reiterate, now is the time to support Leaf in maintaining and compounding on our momentum! We are seeking:
Organisations, competitions, schools, or individuals who can share our upcoming programmes with exceptional students interested in impact - especially in the US.
Collaborators who work with university/early-career students, towards whom we can direct the incredible talent we’ve been identifying and cultivating.
Connections with possible funders to support our work and the increasingly interconnected work of our partner program, Non-Trivial.
This one-page summary provides a quick reference. This full data document has everything. Comments and questions are very welcome, either here or at sam@leaf.courses and jonah@leaf.courses.
Hi Sam, I know an organisation who work very closely with school and university/early-career students in Bulgaria, if that's any use?