.png)
7 hours ago
1
You’re not just a manager at a startup. You’re a therapist, firefighter, recruiter, part-time engineer. And somehow, an island of stability in a sea of chaos.
You’re not in control. Far from it. You just project calm while quietly questioning your life choices. But when you stay calm, others borrow your confidence until they find their own.
Leadership at a startup means helping people move forward through uncertainty: no product/market fit, a half-formed culture, too many ideas, too few people.
Here’s how to stay sane while pretending to know what you’re doing. Most days.
Startups are messy by design. No-one knows what the product is yet, let alone who wants to buy it.
Processes don’t exist. Goals are vague or missing entirely.
At least your priorities are clear: Everything is super-urgent. The critical feature, supporting the new triallists, budget planning, Slack that just won’t stop pinging.
And of course, Bob’s emotional crisis.
Your role, somehow, is to help your people move forward anyway.
It won’t be perfect. Just build enough to learn something, to take one step closer to clarity, then do it all again.
If you can do that without losing your marbles, you’re already ahead of the game.
You’re on a raft barrelling through whitewater rapids. You can’t stop, you can’t go backwards. But you do have a paddle, and with that you can steer, adjust course, maybe avoid a few rocks.
Control in a startup is mostly theatre. The real power is in establishing a calm rhythm: a few rituals that keep people connected and moving. Predictability in small things builds tolerance for chaos everywhere else.
Keep the crew rowing in the same direction while the fog of uncertainty clears. That’s as close to true leadership as it gets.
Once you accept you can’t control the river, your next job is to help everyone keep their head above water.
Engineers don’t like uncertainty. It’s in their DNA. When everything feels vague and undefined, they can slip into analysis paralysis, unable to progress on anything. Does not compute.
This is where you can have a huge impact. Start small. Help your team fill their Certainty Buckets: truths and unknowns. Help them understand that by addressing the truths, the unknowns will come into focus. It’s only by trying something that the fog begins to lift.
Treat every decision like a hypothesis. “We think this will help because…” Then test, observe, and adjust. Don’t waste weeks perfecting a plan that reality will shred in a day.
The trick is to make progress visible. Write things down. Share learnings. Build trust through evidence, not assumption.
Rigour beats bravado. “I don’t know yet” is a valid strategy if it’s followed by “but here’s how we’ll find out.”
At one startup, a new team was formed to improve customer experience. The only problem? We didn’t have any customers. “Where is our signal coming from?”
Seeing the disquiet building, we roleplayed as new customers to test our own assumptions. It was enough to break the team out of their feedback-craving doom-spiral.
When real customers finally arrived, we were wrong about many things. But we were match-fit and learned fast.
It’s increasingly common, and healthy, for Engineering Managers to be hands-on. It keeps them closer to the action and the developer experience.
But that’s not the same as being the hero. A good EM gets their hands dirty to learn, assist and buy breathing space for others. Your code shouldn’t be on the critical path. Someone will always need you mid-merge for something on fire.
Here’s what ‘hands-on’ looked like for me: useful, unglamorous work that made life easier for everyone. Things like:
Building dashboards in Grafana to help inform the team (and myself!),
Adding alerting,
Improving test coverage on a legacy system,
Fixing a slow build,
Adding Kubernetes auto-scaling,
Fielding support requests,
Writing documentation,
Wrangling data.
You build a lot of respect by getting stuck in.
The hardest part is stepping back, watching someone else solve a problem you could fix faster. Your job is to amplify, not output.
In a startup, temperament matters as much as talent. You need people who can function without a spec, who don’t sulk when priorities change, and who can build with a loose brief and a tighter budget.
Look for curiosity. The ones who ask “why” before “how” will save you weeks of wasted effort.
Forget the “startup rockstar” myth. What you want is a dependable grown-up. Someone who doesn’t flinch at ambiguity and can laugh when the plan combusts.
It takes a special sort of person to be a startup CEO. I mostly mean that as a compliment. Mostly.
It’s probably hard when you wake up every day with 5 new ideas and the energy of a small hurricane.
That’s where you come in. You are the Idea Firehose Shield, to absorb, deflect, and occasionally channel all that chaos into something useful.
Say ‘not yet’ often enough and you start to sound visionary instead of obstructive.
Protect your team. Without that shield, context switching becomes the default operating model. Nothing gets finished. Morale evaporates.
At one startup, I learned that language was effective at reframing the firehose of ideas. Use the vocabulary of a CEO, words like “investment”, “risk”, “Net Revenue Retention”. This quickly brings a hand-wavy idea slamming down to earth. They suddenly understand the cost of changing direction and you can get on with what was previously agreed.
Remember: it’s better to finish one useful thing than start five brilliant ones.
Engineering Management is not for the faint-hearted. That makes us expensive, and often a luxury for startups. As teams grow, we can quickly find ourselves spread thin.
In one role, I was managing 17 developers across four teams. Let’s be honest, that’s not ideal.
But this is a startup, right? We find a way.
You delegate where you can. But the real trick is what I call the Swinging Spotlight: a targeted approach that shines on the darkest areas, the team where support is needed most, or the biggest fire is burning.
Of course, this means less support for others. Did I tell you about when I had to write annual reviews for people I didn’t really know (yet)? That smarted. You’ll feel guilty about who’s in the dark. You should. It means you care. Just don’t mistake guilt for failure.
Once you’ve helped one team out their hole, the spotlight swings to the next most critical area.
Management in a startup is triage. The spotlight moves. The art is in moving it fairly, explaining why. Make sure everyone gets a turn in the light, but pace yourself, before you burn out too.
The modern startup EM has a secret weapon: automation that mostly works… except when it doesn’t.
Let AI write your meeting notes, summarise retros, or generate draft job specs. Anything that feels like admin, not leadership.
But exercise caution around privacy and accuracy. When I tried an AI notetaker, it credited the candidate with half my own sentences. Great for their chances, bad for basic truth.
I draw the line at using an LLM to write someone’s annual review. People deserve more than that. However, using a private internal model to digest the 78 pages of 1:1 notes I somehow accumulated? That’s a no-brainer.
Every repetitive task you offload is a sliver of sanity regained. And the fewer chores you have left, the more bandwidth you have for the only part of the job that can’t be automated: people.
AI can draft words. You still have to mean them.
Startups rarely end neatly.
You can’t fix the chaos but you can survive it.
The real skill is composure, staying calm when everything’s on fire, and helping others do the same.
If you’ve made it this far, you’re probably doing fine. You care enough to wonder if you’re getting it wrong. The raft’s still floating. No one drowned this week. That’s a win. Try again tomorrow.
So stay calm and keep paddling.
If this piece hit close to home, you might like my Engineering Manager OS, a Notion template built to give managers the structure I wish I’d had when I started.
