Kartik Chugh · September 25, 2026
In Q2 2025 we took on 9 small retainers in 11 weeks because each one looked easy, and by the end of the quarter our delivery team was working over 6 weeks of sustained overtime while total profit had barely moved. Revenue was up about a third. Margin was down. Nobody had made a bad decision at any individual point, which is what made it hard to see coming.
The mistake was not taking small clients. It was assuming that cost scales with account size, and it does not.
What we thought we were doing
The logic was reasonable. A small retainer is faster to sell, the buyer is usually the founder so there is no committee, and the delivery scope is narrower. We had capacity, the pipeline was full of them, and each one on its own was clearly worth taking.
We tracked the obvious things in HubSpot: contract value, close rate, time to close. All three looked good. What we did not track, and had no field for, was the fixed cost of simply having a client at all.
Where the cost actually was
Every client, regardless of size, consumes a roughly constant amount of certain things. A kickoff. A shared channel in Slack that somebody has to read. A monthly report. A recurring call. An invoice cycle. Access requests and credentials. A quarterly planning conversation. None of that scales down when the retainer does.
When we finally measured it, over 6 weeks of tracked time across 14 accounts, the fixed overhead per client per month was almost identical whether the retainer was small or 5 times larger. On our largest accounts that overhead was a rounding error. On the smallest ones it was consuming most of the margin before any actual work happened.
The compounding problem was attention rather than hours. 9 additional shared channels is 9 additional streams that senior people feel obliged to monitor, and that cost lands entirely on the people who are also responsible for the work that differentiates us. Our strategists were spending their mornings on coordination and their afternoons on the thing clients were actually paying for, and the ratio kept moving in the wrong direction.
The moment it became visible
A strategist asked in a Monday meeting how many clients she was on. The answer was 11. Two years earlier the same role carried 4.
Nobody had decided that. It had happened one reasonable yes at a time, and every one of those yeses had been evaluated on whether we could deliver the work, never on whether we could carry another relationship. We had a capacity model for hours and no capacity model for accounts, so the constraint that was actually binding was invisible to the process that kept adding to it.
We did not understand at the time that these are different resources. Hours are elastic in the short term, which is exactly why they mask the problem: a team absorbs the first few extra accounts by working harder, and the numbers stay fine until they suddenly do not.
The lag is what makes it dangerous. We reconstructed the quarter afterwards and the overtime did not appear until week 7, roughly 5 weeks after the decisions that caused it. By then we had signed 3 more. Any feedback loop where the cost of a decision arrives more than a month after the decision will keep producing that decision, because the person approving the ninth account is looking at data generated when we had five.
What we changed
First, we put a hard cap on accounts per strategist rather than on hours. It is 6. It is somewhat arbitrary and that is fine, because the value is in having a number that forces a conversation before a tenth account arrives rather than after.
Second, we introduced a floor. Below a certain monthly value we do not take a retainer, and we say why. It cost us 2 deals in the following quarter and both of them were deals that would have lost money once fixed overhead was counted honestly.
Third, and this is the one I would recommend to anyone running a small firm, we now price the overhead explicitly. Our smallest tier has a lighter service wrapper: reporting is written rather than presented, there is no standing call, and communication is batched into one weekly summary instead of an always-open channel. Clients on that tier were told plainly what they were getting and almost none objected, because what they wanted was the work.
That third change did more for margin than the floor did. The clients were never the problem. The service wrapper we were applying uniformly regardless of price was the problem.
I want to be fair to the original instinct, because it was not wrong. Small accounts genuinely are easier to win, they diversify revenue away from a handful of large relationships, and 2 of the 9 have since grown into our mid tier. The error was not strategic. It was that we ran a growth push with a cost model that only counted the variable half.
What I would tell another owner
Count your fixed cost per client before you take the next small one, and count it in attention rather than only in hours. The specific test is to ask how many client relationships one senior person is holding, and compare it to what that number was 2 years ago. If it has roughly doubled and nobody decided that, your growth is being funded by capacity you have not priced.
The other thing I would say is that this failure is invisible in every standard report. Revenue rose. Utilisation rose. Close rate held. Every dashboard we had said the quarter went well, and the only signal that anything was wrong was that experienced people were tired and a strategist asked a question in a meeting. Our approach to what actually drives cost per outcome is in our cost per qualified lead breakdown.
We kept 7 of the 9 accounts. They are profitable now, on a different service wrapper, which is the outcome I wish we had designed deliberately instead of arriving at through a bad quarter.