Replacing a developer who is not working out is one of the most avoided decisions in a startup, and the avoidance is expensive. Every week you keep a poor fit in a key seat, the roadmap slips, the rest of the team compensates, and the problem you already know about compounds. The hard part is rarely spotting it. The hard part is acting before you have wasted another quarter hoping it turns around.
Key Takeaways
- Diagnose first: is it a skills gap, an expectations gap, or a genuine mismatch?
- Give one clear, time-boxed chance to improve, with specific and measurable goals.
- Line up the replacement before you remove the person, so the seat is not empty.
- A pre-vetted bench turns a months-long re-hire into a 72-hour shortlist.
First, Diagnose the Real Problem
Before you replace anyone, be honest about why they are struggling, because the fix depends on the cause. Sometimes it is a skills gap the person can close with support. Sometimes it is an expectations gap, where nobody actually told them what "good" looked like, which is a management failure wearing a performance costume. And sometimes it is a genuine mismatch of level or role that no amount of coaching fixes. The first two are worth one real attempt to repair. The third is not, and dragging it out helps no one, least of all the engineer.
Give One Clear Chance, Then Decide
If the problem looks fixable, set a short, specific improvement window: what needs to change, how you will measure it, and by when. Vague feedback ("be more proactive") sets everyone up to fail. Concrete goals ("own the payments migration end to end, with a working staging deploy in three weeks") give a real shot and a clean read. If it works, you kept someone. If it does not, you now have a clear, documented basis to move on, and you gave the person a fair chance on the way.
| Cause | Right response |
|---|---|
| Skills gap | Support + a time-boxed goal |
| Expectations gap | Clarify "good", then re-measure |
| Level/role mismatch | Replace, do not coach |
| Effort/attitude | One direct conversation, then decide |
A Concrete Version
A founder kept a mid-level engineer in a senior backend seat for five months, telling himself the ramp was just slow. Reviews dragged, two features shipped late, and a stronger teammate quietly started redoing the work. When he finally set a concrete three-week goal, the gap was obvious inside one, because the problem was level, not effort. He replaced the seat with a vetted senior from a bench, had a shortlist in seventy-two hours, and the new hire cleared the backlog in a month. The lesson was that five months of hoping cost more than the swap ever would have, not that the first engineer was bad.
The Honest Counterpoint
Moving too fast is its own mistake. If you replace an engineer the moment they struggle, you will churn through good people who were three weeks from clicking, poison your team's sense of safety, and teach everyone that a hard month means the exit. Most strong hires have a rough patch. The judgment is separating a rough patch from a real mismatch, and you do that with a clear goal and an honest look at the cause, not with a snap reaction to one bad sprint. Fast is right for a genuine mismatch and wrong for a normal ramp.
Frequently Asked Questions
How long should I wait before replacing someone?
Long enough to diagnose the cause and run one clear, time-boxed improvement window, usually two to four weeks. Longer than that on a genuine mismatch is just delay.
Should I have the replacement lined up first?
Yes. An empty seat costs you roadmap immediately. A pre-vetted bench lets you have a qualified shortlist ready before you make the change, so there is no gap.
What if I am not sure it is really their fault?
Then it probably is not only their fault. Clarify what "good" looks like, measure against it, and you will get a clean answer either way.
The Bottom Line
The expensive version of this decision is the slow one. Diagnose the real cause, give one fair and specific chance, and if it is a genuine mismatch, move, with the replacement already lined up so the seat never sits empty. For the cost of getting the original hire wrong, see the real cost of a bad engineering hire, and for hiring again after a miss, how to hire a developer after a bad experience. See available engineers.
Roberto Espinoza is CEO of Ruzora, which helps US startups hire pre-vetted senior LATAM engineers, with a vetted shortlist in 72 hours. See available engineers.
