We doubled the team. Why are we slower?
A startup grew from 5 to 26 engineers in under a year, and throughput stayed flat. Here's what was behind it, why AI raises the stakes, and how to find your own team's bottleneck.
By Paul Hepworth, author of The Compounding Team
You doubled the team, and somehow you're shipping slower. It's one of the most common things founders ask me, and it's frustrating because it feels backwards. More people should mean more output.
It usually doesn't, at least not on its own.
What happened when a team went from 5 to 26 engineers
I joined a startup that had grown from about five engineers to 26 in less than a year. On paper it looked great. In practice, here's how work actually flowed:
- The managers designed the solutions with product and design, without their teams.
- The engineers, mostly junior, waited to be handed requirements and told how to build them.
- Everyone was comfortable, except the founders. Costs had doubled and throughput was flat.
Nobody was doing anything wrong on purpose. The managers were protecting their teams. The engineers were doing what they were asked. But every piece of work had to pass through a manager twice: once on the way in, and again every time someone had a question.
Brokers vs. multipliers
So we ran an experiment. We changed how the managers thought about their job.
| A broker manager… | A multiplier manager… |
|---|---|
| Sits between the team and product, design, and customers | Connects the team directly to product, design, and customers |
| Designs the solution and hands it down | Helps the team understand the problem and design the solution |
| Answers every question | Helps people unblock themselves |
| Adds a handoff to every piece of work | Removes handoffs |
Within four to six weeks, the gaps in skill and will across the teams became very visible: who could own their work end to end, and who needed help to get there. That was uncomfortable, and it was exactly what we needed. When things are visible, they're fixable.
Why AI raises the stakes
You might be thinking: I'm not planning to double my team. We're using AI.
That may be true. But there's never been a time when it mattered more for engineers to be close to the customer. AI multiplies what an engineer already understands. An engineer who knows what the customer needs can use it to build the right thing fast. An engineer who's waiting for instructions just gets to the wrong thing sooner.
Engineers need to be mini product managers now, not order-takers.
Your bottleneck might be different
In this story, the bottleneck was how managers saw their role. Yours might be ramp-up time, tech debt, unclear goals, or decisions stuck with a few people. Before you fix anything, find it.
- Start with what people see as slow. Then peel back the onion. Where is work harder than it needs to be? Look at the data, not just opinions.
- Map how work flows, from idea to in customers' hands. Follow the information, the decisions (and who makes them), and the actions.
- Look for the waste: handoffs, waiting, and rework.
- If decisions are the slow part, ask whether your engineers are fully ramped up, and whether you're missing systems like clear goals, written standards, or architecture principles that would make many of those decisions unnecessary.
Try this week
Pick one piece of work that felt slow. Walk it back from shipped to idea, and mark every handoff and every wait. The bottleneck is usually hiding there.
If you'd like a second opinion, send me what you find.
Go deeper
- In the book
- Chapter 1: How a company compoundsStart with "The Five Focusing Steps" and "The Bottleneck Moves."
- Chapter 10: Team processes"The Theory of Constraints as a North Star" and "Lean Principles and Organizational Waste" cover handoffs, waiting, and mapping how work flows.
- Chapter 11: Decision-making and ownership"Directly Responsible Individual (DRI)" covers who owns a decision, and why it shouldn't always be the manager.
- In your team's report
- The free assessment scores Focus & priorities, Decision-making & ownership and Role clarity & accountability against your stage. See where your team stands →
Questions
- Why does adding engineers sometimes make a team slower?
- New people take time to ramp up, and every person you add creates more handoffs and more decisions waiting on someone. If the way work flows doesn't change as the team grows, the extra people mostly add waiting.
- What's the difference between a broker manager and a multiplier manager?
- A broker sits between the team and product, design, and customers, and passes requirements along. A multiplier helps everyone on the team understand what customers need and work directly with product and design, so the team can make good calls and unblock itself.
- How do I find my team's bottleneck?
- Start with what people see as slow. Pick one piece of work that felt slow and walk it back from shipped to idea. Mark every handoff, every wait, and every decision someone had to wait on. The bottleneck is usually hiding there.
- Does this still matter if we're using AI instead of hiring?
- More than ever. AI multiplies what an engineer already understands. Engineers who know the customer's needs get far more out of it than engineers who are waiting to be told what to build.
Transcript
So you doubled your team size, but not your throughput. Uh-oh. Welcome to What's Holding You Back, a series where I share some insights on questions and situations that you submit. The question from this week is from a founder: why is my team moving slower as we hire more people?
It reminds me of a story from when I joined a team that had recently grown from about five engineers to 26 engineers in less than a year. I observed that the managers designed the solutions with product and design, without their teams. And the junior engineers on those teams were all too happy to be told what to do, to be handed requirements, and to be told how to build it. Everyone was comfortable with this, except for the founders. They saw that costs doubled and throughput was flat.
I sat down with the engineering managers, and we decided we would run an experiment. What if we changed how the managers thought about their work? No longer being brokers of information, protecting the team from the customer and from design and product, but being multipliers. How could they help every person on the team understand customers' needs, work more closely with product and design to come up with the best solutions, unblock themselves, move forward quickly, and remove handoffs?
Four to six weeks later, the gaps in skill and will across the teams became very visible. And when things are visible, they're fixable.
Now you may be thinking, I have no plans to double my team size because I'm using AI. And that may be true. But there's never been a time when it's more crucial for the engineers building your software, using the power of AI, to be closer to customers' needs. They're mini product managers now, not order takers.
So in this particular story, the team needed to reevaluate what a manager was and how they supported the team. But that may not be the situation for you. What's your bottleneck, and how is it holding you back?
My suggestion: start with what you see as slow. Peel the onion back, look at the data. What's harder than it should be? Make yourself intimately familiar with everything that happens between the idea and delivering it to customers. What information flows? What decisions are made? Who makes those decisions? What handoffs are made? Where's the delay? Pick one slow piece of work and walk it all the way back from shipped. Review the data and how it went, understand it, identify where the bottlenecks are, and fix them.
I hope this was thought-provoking and helpful. That's this week's What's Holding You Back. Send me yours: what's holding you back? The link is in the comments, and anonymous is fine. I can't wait to hear what you have to say.
What's holding your team back?
Send me your situation. Every week I answer one: what's usually behind it, and one first move. Anonymous is fine.
Send yours →