Graduate programmes tend to follow a shape. Year one: a strong cohort, high energy, visible wins. Year two: attrition among the first cohort, a second cohort that gets less attention, and a quiet conversation about whether the programme is worth it.
The causes are almost always decisions made in year one.
Hire for the second year, not the first
The temptation is to hire graduates for what they can do in month three. The people who make a programme work are the ones still there in month thirty. Screen for the ability to learn in your environment specifically — which means asking about times they were dropped into something unfamiliar, not testing algorithm trivia.
Budget the mentoring properly
Every graduate hire consumes senior time. A rough figure from the teams we work with is four to six hours a week of a senior engineer’s attention for the first quarter. If that is not planned, it is taken — from whoever is kindest, which is how you burn out your best mentors and create your second vacancy.
Name mentors explicitly, count the hours in their capacity, and review it in their performance cycle. Mentoring that is invisible in someone’s objectives is mentoring that stops under pressure.
Give the first project real stakes and a small blast radius
Nothing disengages a graduate faster than three months of make-work. Nothing damages confidence faster than owning something critical too early. The middle ground — genuinely used, reversible if wrong — is where people learn quickly and safely.
Decide what year two looks like before year one starts
Where does a graduate go after eighteen months? If the answer is “we will see”, your strongest people will leave, because they will read the absence of a path as an absence of a future. This does not require a formal ladder; it requires being able to say what the next step is and roughly when.
Do not run it every other year
Alternating cohorts is a false economy. A second cohort arriving a year later has no peers a year ahead of them, which removes the single cheapest support structure a programme has — near-peer help. Smaller and annual beats larger and intermittent.
Where external help fits
Sourcing volume at graduate level is a distinct skill with a distinct calendar — campus timelines, cohort deadlines, offer windows that close all at once. Most in-house teams are structured for continuous hiring rather than seasonal peaks, which is why graduate sourcing is one of the more common things we get asked to run.
