Episode Details
Back to Episodes“From job posting to hire: templates, sourcing campaign, and LLM-resistant tasks” by Romain Barbe🔸
Description
This document explains how Mieux Donner ran its 2026 hiring round: how we decided what to hire for, how we built the offer and the process, the results we got, and what we would tell another organisation doing roughly the same. It is meant to be reused. We also advise you to read the chapter on hiring in “How to Launch a High-Impact Nonprofit”.
Mieux Donner is the French effective giving initiative, incubated through Ambitious Impact (AIM) and Giving What We Can in 2024. We were roughly 2FTE, have directed over €1M to high-impact charities at a giving multiplier of 5 to 6 times and are now looking to expand the team.
I used AI to do some analysis on the application (without applicant data) and to correct my speech-to-text.
A note for applicants: This document is written for people running a hiring process, not for people applying to one. Reading it will probably not help you, and we do not really advise it. Knowing how a process is designed could be useful if you are applying to a government body or a high-earner position, but the process we follow is unlikely to resemble any of those. [...]
---
Outline:
(02:04) 2026. at a glance
(02:19) 1. Deciding what to hire for
(02:24) Budget and contract
(02:38) Why we opened four roles to hire two people
(05:03) Choosing what to test for
(05:18) 2. The offer
(05:21) Open or closed round?
(06:39) A public offer, and why
(08:28) Sourcing
(09:44) Salary
(11:38) Referrals
(12:08) 3. The process
(14:20) Defining the weights
(15:30) Rating scale and calibration
(16:52) Designing questions LLMs fail, and benchmarking against them
(17:37) Making the tasks LLM-resistant
(20:15) Emails, handling, tracking
(21:36) Value personalised feedback
(22:35) Minimum scores for progression
(24:04) Biggest gap: delay to process applications and opening length
(25:11) After the process email
(25:42) 4. The 2026 results, and what we learned
(25:48) The funnel
(26:46) What actually predicted who advanced
(27:06) Biographical data: an open question
(27:29) Where the good candidates came from
(29:21) We changed the relative weight of TestGorilla vs tasks
(29:46) What we learned about the interview
(30:10) AI in the application
(30:59) 5. The time it takes
(31:37) 6. What we would improve
(33:09) What we are unsure
(34:39) Reusing this
---
First published:
June 17th, 2026
---
Narrated by TYPE III AUDIO.