The Bottleneck Test: How to Find the One Thing Capping Your Growth

Your business doesn’t have ten problems. It has one problem and nine annoyances, and learning to tell them apart is worth more than anything you’ll buy this year.

Growth never stalls everywhere at once. It stalls at one point: something specific, usually quiet, that everything else ends up waiting on. Manufacturing people figured this out decades ago with the theory of constraints. A production line moves at the speed of its slowest station, and improving any other station changes nothing except the pile of inventory in front of the slow one. Your business runs on the same physics. It just hides them better.

Why owners fix the wrong thing#

The problem that gets your attention is the loudest one, and the loudest one is almost never the binding one. An overflowing inbox screams at you every morning. A pipeline that’s been quietly soft for two quarters says nothing at all. So the inbox gets the new tool, the assistant, the weekend of cleanup, and revenue doesn’t move, because revenue was never waiting on the inbox.

We wrote in our first post that most AI disappointment comes from automating work that wasn’t the constraint. The line from our notes at the time still holds: automating the wrong thing just gets you more of what wasn’t holding you back. But that raises the question this post answers. How do you find the right thing?

The doubling question#

Here’s the test, and you can run it alone with a coffee in twenty minutes. Ask yourself: if twice as many leads called tomorrow morning, what would break first?

Follow the work through your business and watch where it piles up. There are only four places the answer can land.

Demand. If your honest reaction is “nothing would break, I’d love that,” your constraint is upstream: not enough interested strangers even reach you. The signs are a quiet pipeline, feast-and-famine months, and marketing that happens in bursts when things get scary.

Conversion. If leads already reach you and quietly die, doubling them just doubles the dying. The signs: quotes that go out and never get a follow-up, inquiries answered on day three, a pile of “I’ll get back to you” that nobody got back to.

Cash flow. If more sales would actually make the pressure worse, money is your bind. You’re profitable on paper and tight every single month, because cash arrives sixty days after the work and growth widens that gap instead of closing it.

Capacity. If the first thing to crack is delivery, you’re at the constraint most established businesses hit: work is already being turned away or delayed, and too much of it waits on you or your two key people. Everything piles up in front of the same desks.

Whatever breaks first is your constraint. Not your only imperfection, not your only irritant. The one thing currently deciding the speed of everything else.

One constraint per quarter#

Now the uncomfortable part: this quarter, the constraint gets the money and the hours, and everything else gets maintenance. Improving three areas at once feels responsible and accomplishes roughly nothing, for the same reason speeding up three fast stations doesn’t move the line.

This is also where every AI and delegation decision becomes simple. As we argued last week, nobody actually wants AI; you want specific work delivered finished. The bottleneck test tells you which work: the recurring layer sitting on your constraint. If conversion is the bind, the follow-ups and quote responses are worth delegating this quarter, and the invoice automation you were considering can wait, however satisfying it sounded.

What the test looked like at the school#

At Académie Lavalloise, the 250-student school that runs on Avelle, the diagnosis came before any technology. The constraint was capacity: a small administrative team buried under recurring communications, with enrollment season as the annual breaking point. Doubling inquiries would have snapped delivery, not demand. So the first deployment absorbed exactly that layer, twelve recurring processes, an estimated 35 to 50 hours a week, and the freed capacity went back to the work only humans should do. The order mattered. Name the constraint first; only then decide what gets automated.

Run it this week#

Twenty minutes, one question, four candidate answers. Write down what would break first, then list the recurring work piling up in front of that point. That list is your quarter. If you want to pressure-test your answer with someone who runs this diagnosis every day, a thirty-minute conversation is enough for us to tell you honestly whether you’ve named the right constraint, or whether the loud problem fooled you too.

One thing is capping your growth. It deserves to be found.

---

Matthew Lacopo leads client engagement and content at The Agensee, a managed AI operations firm based in Montréal serving SMBs across North America, which he cofounded with Anthony Lacopo.