Each year in England, around a quarter of the people who qualify as teachers don’t go on to take up a post in a state school. They have invested a year of their lives in becoming qualified. And then they don’t teach. Why? This is one leak in what’s often called the leaky teacher pipeline. The metaphor is useful, but it can hide what’s actually going on. Sometimes the pipeline simply isn’t bringing enough people in at the top. Other times the leak is at the joints, where one stage hands off to the next They don’t all have the same cause. Across two new studies, funded by the Gatsby Foundation, we found leaks at different stages of the pipeline appear to need different fixes. Turning on the tap Teaching in England is a graduate profession, so undergraduates are the pool the next generation of teachers will come from. In the first study, we surveyed over 2,000 undergraduates in the UK and US, showing them pairs of hypothetical jobs that varied on things like salary, working hours, paid leave, social impact, and flexibility, and asking them to pick which they would prefer. By varying these features at random, we could see which ones actually shifted people’s choices. The biggest drivers were extrinsic: pay, hours and paid leave. Social impact mattered too, but less. This held for the subset of undergraduates who said they were considering teaching as well as for those who weren’t. This sits awkwardly with existing research on entry to teaching, which has tended to emphasise altruistic motives. One likely reason for the difference is method. Most existing studies simply ask people why they want to teach, which tends to produce answers that sound good rather than answers that reflect the actual trade-offs people make. The implication is that recruitment campaigns leaning heavily on the mission-driven aspects of teaching may be missing a trick. The extrinsic features of the job, including the relatively generous paid leave teachers in England receive, are highly attractive to undergraduates and worth communicating clearly. The leak between training and teaching If extrinsic rewards are what attract people into training, you might expect them to be what determines who enters teaching after qualifying. But the second study suggests something different is going on at the joint between training and employment. We tracked over 400 trainee teachers across three large initial teacher education providers in England, from the middle of their training year into the following autumn. We asked about their job preferences and how their training experiences had compared with their expectations across ten dimensions of teaching work, including pupil behaviour, mentor support, and lesson planning workload. There are two leading explanations in the literature for why qualified trainees don’t enter teaching. The first is that those who enter training have preferences that turn out, in retrospect, to be a poor match for what teaching offers. The second is “reality shock”, the idea that trainees enter with similar preferences to other graduates, but their experiences during placements diverge sharply from expectations. We found little evidence for either. Trainees who didn’t go on to teach had essentially the same job preferences as those who did. Trainees were shocked by some aspects of training. For example, admin and lesson planning were worse than expected, while relationships with pupils were better. But those shocks didn’t predict whether trainees actually entered the profession. What did? Open-text responses from trainees who had intended to teach but ultimately didn’t pointed to strikingly idiosyncratic reasons: illness, bereavement, house fires, visa issues. Several said they planned to enter teaching a year later. The gap between qualifying and entering may have less to do with teaching itself and more to do with the messy realities of life during the transition. More than one repair job Taken together, these findings suggest that attracting undergraduates into teacher training and getting qualified trainees into a classroom are two different problems. Attracting undergraduates is about making the job competitive against other graduate options, and our evidence suggests pay, hours and leave are what matter most. Getting qualified trainees into a classroom is something else entirely, and may have less to do with teaching itself than with supporting people through a vulnerable transition. The teacher pipeline is a series of connected stages, each with its own pressures and its own leaks. Fixing it means treating it as a system, not a funnel.