Engineering Leadership · Chapter 3 of 9
Growing people, and managing someone out
The two halves of the same skill. Most leaders are competent at one and avoid the other.
Performance work is where leadership stops being pleasant. The interview will probe both directions, and candidates who have only ever been supportive or only ever been decisive both read as incomplete.
Diagnose before you intervene
A sudden drop in someone's output is a signal about the system or the situation until proven otherwise. In practice the causes cluster: an unfamiliar area handed over badly, something happening outside work, a mismatch between the role and what they are good at, or genuine capability limits. Only the last is a performance problem, and it is the least common. Reaching for a plan before diagnosing adds scrutiny to someone already struggling and usually makes it worse.
When it is the right call
- Be clear, early and specific. The cruelty in performance management is ambiguity — months of hints followed by a sudden formal process. If someone is surprised by a formal conversation, that is your failure not theirs.
- Write down what ‘better’ means in terms they can act on, with a date. ‘More proactive’ is not actionable.
- Rule out what is yours to fix first — unclear expectations, missing context, a bad team fit — and say you have. That is what makes the eventual decision defensible to them and to you.
- Then decide, and do not let it drift. A prolonged process is worse for the person than a clear one, and the team already knows.
Growth, which is the larger half
Most of your leverage is in the people who are doing fine. Ask what they want to be able to do in two years, then find work that moves them toward it — and be honest when the organisation cannot offer it, because helping someone leave well is a better outcome than keeping them frustrated. Leaders whose people leave for bigger roles elsewhere and stay in touch have a hiring advantage that compounds for a decade.
Feedback that changes behaviour
Most feedback fails for one of three reasons: it is too late, too general, or aimed at the person rather than the behaviour. All three are fixable and none require a framework.
| Instead of | Say | Why it lands |
|---|---|---|
| "You need to be more proactive" | "In the last two design reviews you waited to be asked before raising the migration risk. Raise it first next time" | Specific, observable, and actionable this week |
| "Your communication could be better" | "The status update didn't say what was blocked. Lead with the blocker and what you need" | Names the artefact and the fix |
| "Great job on the launch" | "You caught the schema issue two days before release — that's the second time you've found something in staging that the tests missed" | Praise that names a pattern reinforces the pattern; generic praise reinforces nothing |
| "You're not at senior level yet" | "Senior here means owning an outcome including when the requirement is wrong. Here are two moments this quarter where that was available and didn't happen" | Converts a judgement into a description of the gap |
Managing someone out, done properly
By the time you are formally managing someone out, the failure has usually already happened — and it is nearly always the same failure: nobody told them clearly, early enough, in words they could act on.
The test: if you asked the person today whether their job was at risk, would they say yes? If the answer is no and you are three weeks from a formal process, that is on you, and it is worth saying so out loud to yourself before you say anything to them.
What a fair process actually requires: the gap stated in specific, observable terms; what ‘fixed’ looks like, concretely; a genuine timeframe; real support rather than a paper plan; and no surprises at any point.
The most common mistake is a plan designed to document rather than to succeed. People know the difference immediately, it destroys trust with everyone watching, and the team always knows.
The kindest version is also the fastest one. Long ambiguous periods are worse for the person than a clear conversation, because they cannot make decisions about their own life while the outcome is undefined.
And be honest about the possibility that the role is wrong rather than the person. A strong engineer failing in a role is sometimes a mis-hire, sometimes a mis-fit, and sometimes a management failure — unclear expectations, an impossible remit, no support. Those look identical from the outside and require completely different responses, and the diagnosis is your job, not theirs.
Check yourself