Most Agency Problems at $3M Trace Back to Delivery Nobody Designed
Most of the problems agency owners bring me trace back to one root: nobody ever designed how client work gets delivered.
It got built on the fly. A client asked, someone figured out how to deliver it, and it snowballed from there. That’s fine until you hit $3M and the cracks start showing up everywhere at once.
When owners describe those cracks, they describe them as separate problems. They talk about them in separate meetings, with separate people, and try separate fixes. Here are the four I hear most.
You can’t get out of delivery
The process only lives in people’s heads, so quality drops the second you hand it off. You stay the safety net on the biggest accounts, so the agency can’t run without you.
Most owners try to fix this with a hire. They bring in a director of client services or an operations lead and expect the owner’s load to move over. It rarely does, because the new person inherits a process that was never written down. They spend their first months learning it from the owner, which keeps the owner in delivery even longer.
Your margin is thin and you can’t say why
Nobody knows what a single deliverable costs to make. Services are priced off what you charged years ago, and the costs have crept up since.
The usual fix here is a finance fix. Tighten up the books, cut a tool or two, push harder on utilization. Those can help at the edges, but the P&L can’t tell you which deliverable is losing money. That answer lives in how the deliverable gets made, which is a delivery question.
AI isn’t helping the way everyone promised
It got rolled out where it was easy, not where it mattered. The hours it saves go to the client as lower prices or extra work nobody bills, or to the team as slack that fills with meetings.
Those hours only reach you if someone redesigns the steps and resets the price. When AI gets added to a process nobody mapped, the time it saves has nowhere specific to go, so it spreads out and disappears. The team feels a little less rushed for a month, and the margin doesn’t move.
A buyer would pay less for the business than you think
Thin margins and a founder the business can’t lose both drag the price down.
Owners usually find this one out late, when they start talking to a broker or a buyer and hear the number. The fix most people reach for is cleaning up the financials before a sale. But a buyer is pricing the risk that the business falls apart without you, and no amount of financial cleanup removes that risk if the delivery still runs through you.
Why they’re one problem
All four come from one source. Delivery was never designed, so you’re running a $3-5M business on a delivery model built for a much smaller one.
Look at how they connect. The process lives in people’s heads, so you can’t hand it off, so you stay in delivery. Because nobody mapped the steps, nobody knows what a deliverable costs, so the price drifts and the margin thins. With no map, AI lands wherever it’s easiest and the savings leak out. And a business with thin margins that depends on its founder is worth less to a buyer.
Pull on any one of those and you end up in the same place, at the question of how a deliverable gets made and who makes it.
“These are different problems with different owners”
That’s how most leadership teams treat them, and it makes sense on paper. The founder owns getting out of delivery. Finance owns margin. Whoever’s most excited about AI owns AI. The exit is a someday problem.
The trouble is that each owner works on their piece without touching the shared cause. Finance can’t fix margin without changing how deliverables get made. The AI champion can’t capture savings without changing the process and the price. The founder can’t step back until the process is written down. Everyone ends up waiting on a piece of work nobody owns.
That’s why the problems seem to resist fixing. They’re getting worked on from four directions and none of them reaches the root.
Design it, one service at a time
The fix is to design delivery, one service at a time. Map the steps, cost them, cut the leaks, put AI where it belongs, and reset the price.
Starting with one service keeps it small enough to finish while the team is still delivering for clients. Pick the one that worries you most. Sit with the people who deliver it, write down every step with an owner and an hour estimate, and compare that to what the time data says. Fix the rework, move the mechanical steps to AI, decide where those hours go, and set the price based on what the service costs to make.
When you do that, the four symptoms start clearing up together. The process is written down, so you can hand it off. You know what it costs, so you can price it. AI has a defined place to go, so the savings land somewhere you can see. And the business gets easier to run without you, which is the thing a buyer is paying for.
The symptoms clear up together because they were never separate problems.