Skip to main content
Leading Through a Hypergrowth Phase — Anselm Fowel
Leadership

Leading Through a Hypergrowth Phase

10 min read
751 views
Share:

We went from 40 engineers to 140 in about fourteen months. I have the calendar invites to prove it, and the scar tissue too. Somewhere around headcount 90 I realized that almost every management instinct I trusted at 40 people was quietly working against me, and nobody sends you a memo about that. The org just starts groaning in ways you can't immediately name.

Leading Through a Hypergrowth Phase
Leading Through a Hypergrowth Phase

Hypergrowth is romanticized in a way that annoys me. The board deck shows a line going up and to the right. What it doesn't show is the on-call engineer who joined three weeks ago being paged for a payment-reconciliation job she has never seen, at 2am, with no runbook, because the person who wrote it left and nobody had time to document it. That's what growth actually feels like from the inside. Here's what I've learned about leading through it without setting the house on fire.

Hiring is not the hard part

Everyone treats a hypergrowth phase as a recruiting problem. Fill the seats, hit the plan. But hiring 100 people is the easy 20% of the work. The hard 80% is that every single one of those people needs context, and context does not scale the way headcount does. When I had 40 engineers, most of the important knowledge lived in about eight people's heads, and that was fine, because those eight people talked to each other constantly. At 140 that same informal network becomes a bottleneck and then a liability.

The failure mode is subtle. Velocity does not drop off a cliff; it erodes. A change that used to take two days starts taking five, because the new engineer doesn't know that the fraud-scoring service has an undocumented dependency on a nightly batch that must finish before 6am. Nobody logs that as a problem. It just shows up as a slightly worse quarter, and then another one, and by the time it's obvious in the metrics you're two quarters deep into a culture that shrugs at things being slow.

My rule now: for every cohort of hires, I want a named owner whose explicit job for that quarter is making those people productive, not shipping features. That is an unpopular thing to say to a VP of Product who is counting on the new capacity. I say it anyway.

The org chart is a forcing function, not a diagram

I used to think reorgs were something you did reluctantly, when things broke. During hypergrowth I flipped that entirely. The org chart is one of the few levers a CTO has that actually reshapes behavior, and if you wait until it's obviously broken you've already paid a quarter of tax. Conway's Law is not a cute observation; it's a physical constraint. Your architecture will come to mirror your team boundaries whether you plan it or not.

So I started drawing the team boundaries I wanted the architecture to have, then staffing to them, rather than the reverse. When we needed our ledger and our payments-orchestration layer to have clean, versioned APIs between them, the fastest way to get there was to put them in separate teams with separate on-call rotations and make them negotiate the contract. Suddenly the interface got documented, because both sides had a reason to care. Structure did the work that three architecture review meetings had failed to do.

The trade-off is real and I won't pretend otherwise. Every reorg costs you weeks of unsettled people wondering if they still report to the manager they liked. Do it too often and you breed cynicism. I try to hold major boundary changes to roughly twice a year and communicate the why with almost embarrassing repetition.

Promote managers slower than you want to

This is the mistake I regret most from our fastest stretch. When you're doubling, you have an enormous number of new reports and not enough managers, so the temptation is to take your best senior engineer and hand them six direct reports on a Friday. I did this maybe five times in a year. Two of those worked out. The other three were painful for everyone, including me, because I had set good engineers up to fail at a completely different job with almost no support.

Managing during hypergrowth is harder than managing in steady state, not easier, and yet we throw the least experienced managers straight into it. A first-time manager who inherits a stable team of five can learn on the job. A first-time manager who has to hire four people, integrate them, and keep the on-call rotation from collapsing is being asked to run before they can walk.

The single highest-leverage thing I did in that period was hire two seasoned engineering managers from outside who had scaled an org before. They cost more and they slowed my headcount plan by a month. They also quietly prevented at least three resignations I only heard about later.

If I could rerun the tape, I would keep more strong engineers as strong engineers, pay them like the assets they are, and be far more deliberate about who I put into management and when.

What actually breaks first

People assume the systems fall over first. In my experience the systems are surprisingly resilient; it's the human coordination that goes first. Here is roughly the order things broke for us, and I've since heard the same sequence from peers at other fintechs:

Enjoying this article?

Get more like it in your inbox — practical engineering leadership, fintech, and AI. No spam, unsubscribe anytime.

  • Onboarding. The informal buddy system that worked at 40 collapses because the buddies are drowning.
  • Decision-making. Choices that one person used to make in a hallway now need three teams in a room, and no one owns the call.
  • Incident response. More services, more surface area, and half the responders have never seen the system in a bad state.
  • Documentation. It was always thin; now the gap between what's written and what's true becomes dangerous.
  • The systems themselves, eventually, usually because of a change made by someone who lacked the context above.

Notice that the technology is last on that list. If you spend your hypergrowth phase mostly worrying about whether the database will scale, you are optimizing the thing least likely to hurt you. The database has a vendor and a knob. Your onboarding does not.

Protect the boring work

In a regulated environment, the boring work is the work that keeps you in business. Reconciliation, audit logging, access reviews, the controls your auditors will ask about in eleven months. During hypergrowth every incentive pushes you to defer this, because it doesn't show up in the demo and the new hires all want to build the shiny thing. I have watched teams quietly accumulate a compliance debt that was far more expensive than any technical debt they were tracking.

I now ring-fence a fixed fraction of engineering capacity for this category and defend it like it's payroll, because functionally it is. Roughly 15 to 20 percent, depending on where we are in the audit cycle. When a product leader pushes back, and they always do, I remind them that a failed control finding can freeze a product launch for a quarter. That tends to reframe the conversation faster than any principled speech about engineering hygiene.

The thing nobody tells you is that the boring work is also where your newest engineers can learn the domain safely. Handing a new hire the access-review tooling is a gentler introduction to how money actually moves through your system than dropping them into the payments hot path on day four.

Communication scales worse than anything else

At 40 people I could stand up in an all-hands and genuinely believe the message landed. At 140, the same talk reaches maybe half the room in the way I intended, and a distorted version reaches the rest through hallway retellings. I underestimated this badly. I thought if I said something clearly once, it was communicated. It was not.

What works is repetition to the point where I feel slightly ridiculous, plus writing things down so the message survives the retelling. I write a short weekly note now, maybe 400 words, on what I'm thinking about and why we're making the calls we're making. It is not polished. It is consistent. The consistency matters more than the polish, because in a fast-growing org the absence of information gets filled with anxiety, and anxious engineers make worse decisions and update their resumes.

The corollary is that you have to listen at scale too, and skip-levels stop being enough when the org triples. I lean on a short quarterly engagement survey and, more importantly, on the two or three long-tenured engineers who will tell me the unvarnished truth over coffee. Guard those relationships. They are your early-warning system.

Say no to most of it

Hypergrowth generates an absurd volume of opportunity. New partnerships, new markets, new product lines, all arriving faster than you can staff them. The strategic error is treating capacity as if it will catch up to ambition. It won't, not during the growth phase, because every new hire is a net drain on the existing team's time before they're a contributor.

I got much more comfortable saying no, and more specifically saying "not now." We turned down a sizable partner integration that would have opened a new vertical because taking it would have meant pulling four engineers off the ledger migration we absolutely had to finish. On paper it looked like leaving money on the table. In practice, finishing the migration cleanly let us onboard the next three partners in weeks instead of quarters. Focus compounds; scatter does not.

My personal test for any new commitment during that period was simple: if we say yes to this, what specifically stops, and who has to be told. If I couldn't answer both parts, I wasn't allowed to say yes.

You will change as a leader, or you'll be the bottleneck

The uncomfortable truth is that the CTO who was excellent at 40 people can be actively harmful at 140 if they don't change what they do all day. I loved being in the code. For a while I kept reviewing pull requests in areas I no longer understood well, and I was slowing good engineers down while telling myself I was staying close to the work. I was flattering my own ego and calling it engagement.

The shift is from being the person who has the answers to being the person who makes sure the right people have the context and the authority to find the answers themselves. That is a genuinely different job, and some days I still miss the old one. But an org that has to route decisions through you cannot grow past the throughput of your calendar, and your calendar does not double.

Anselm Fowel, CTO and fintech architect
Anselm Fowel — CTO & fintech architect

Conclusion

If there's one thing I'd tell a leader heading into this, it's that hypergrowth doesn't reward the person with the most energy. It rewards the person willing to be a little unpopular in the short term to protect the things that don't show up in this quarter's numbers. The onboarding nobody wants to own, the compliance work nobody wants to staff, the manager promotion you decline to make. Those choices feel like friction against the momentum, and they are the entire job. Grow the line on the chart, sure. Just make sure there's still a company underneath it when the line finally levels off, because it always does, and that's when everything you skipped comes due at once.

Enjoyed this article? Share it with others!

Share:

Get new posts in your inbox

Occasional, practical notes on engineering leadership, fintech, and building with AI. No spam, unsubscribe anytime.

Comments (0)

Leave a Comment

Comments are moderated and will appear after review.

No comments yet. Be the first to comment!

About the author

Anselm Fowel

Anselm Fowel

Chief Technology Officer & fintech architect. 16+ years leading engineering across AlliancePay, Mondu, Transalliance, Global Accelerex, and Fidelity Bank — writing here about engineering leadership, fintech architecture, and AI in production.

Read next