What Actually Predicts Success in the First 90 Days
A strong interview process doesn't guarantee a strong start. When we follow up with clients after a placement, a consistent set of behaviours separates the engineers who ramp up quickly from those who struggle, and it has less to do with raw technical ability than most people assume.
Does interview performance reliably predict how someone performs once they're actually in the role?
Not as reliably as either side usually expects. Plenty of candidates who interview extremely well take much longer than expected to become genuinely productive, while others who interview solidly, but not spectacularly, ramp up fast once they're actually working inside the real systems and team. What separates the two groups isn't raw ability, it's a specific set of early behaviours.
What's the clearest early signal that a new hire is going to succeed?
How quickly they get real, working code into production. Engineers who succeed tend to ship their first genuine production change within their first couple of weeks, and settle into a rhythm of regular, frequent contributions from early on. Engineers who struggle tend to spend an extended period reading documentation and waiting for explicit permission or direction before making a real contribution, treating the ambiguity of a new environment as something to wait out rather than work through.
Does physically engaging with the trading desk actually matter, or is that just company culture talk?
It matters genuinely, not just as a culture signal. Successful new hires make a point of walking over to the desk, spending time with quantitative researchers and traders, and building a real understanding of how the systems they're working on actually affect trading outcomes day to day. Engineers who stay isolated behind their own screen, technically capable but disconnected from the desk, consistently take longer to build the trust and context that lets them operate independently.
What happens when something breaks, is that a good test of a new hire?
It's one of the clearest tests there is. The ability to jump into a live production issue, trace the problem across unfamiliar parts of the system, and work toward a root cause under real time pressure, without immediately escalating to someone more senior, is one of the strongest signals of genuine readiness. New hires who can do this credibly within their first few months tend to be trusted with real ownership much faster than those who can't.
If interviews don't fully predict this, what should clients and candidates actually focus on?
For clients, it's worth being explicit early about the pace of ownership expected, so a new hire isn't quietly waiting for a level of structure and permission that was never coming. For candidates, the honest advice is to treat the first few weeks as a deliberate push to contribute something real quickly, engage directly with the desk rather than staying heads-down, and treat an early production issue as an opportunity to build trust rather than something to avoid or escalate immediately.
