By Diosh — Founder, AHAeCommerce | eCommerce decision intelligence for $50K–$5M GMV operators
This is a mistake piece for operators who have been carrying an underperforming or culturally-misaligned hire for months and keep telling themselves to give it one more cycle. The mistake is not the bad hire — every operator makes those. The mistake is the tolerance: the slow, defensible-sounding decision to wait, coach, and hope while the cost compounds in places your P&L does not show. By the time the average operator acts, they are 3-6 months late, and that delay is more expensive than the severance they were trying to avoid. This article gives you a test to know when you already have the answer, and a process to act on it cleanly.
The Tolerated Mis-Hire Is the Most Expensive Line Item You Don't Track
Your underperformer's salary is the visible cost. It is also the smallest one. A $58K customer-experience lead who is at 60% of the bar is not costing you $58K of waste — they are costing you the fully-loaded drag on everyone who routes work around them.
The Society for Human Resource Management (SHRM) has long estimated that replacing an employee costs between 50% and 200% of that person's annual salary once you count recruiting, onboarding, lost productivity, and ramp time. Operators read that number and conclude firing is expensive, so I'll wait. That is the exact inversion that traps them. The replacement cost is a one-time, bounded expense you control the timing of. The tolerance cost is an open-ended bleed you do not control at all — it runs every single day the seat is occupied by someone below the line.
A $1.4M GMV supplements brand kept an operations coordinator nine months past the point the founder privately knew it wasn't working. The math the founder eventually ran: roughly $43K in salary over those nine months, plus an estimated 6-8 hours per week of the founder's own time absorbing escalations and re-checking work — call it 270 founder-hours that should have gone to merchandising and supplier negotiation. At even a conservative internal value on founder time, the waiting cost more than two full severance packages. The salary was never the expensive part.
When you are sizing what a seat actually costs you, the framing in the contractor-vs-employee math applies in reverse here: the question is not "what do they cost me to keep" but "what is the fully-loaded cost of the output I am not getting." A seat below the bar has negative leverage. It does not just fail to add — it subtracts from everyone connected to it.
The Hidden Cost Nobody Puts on the Spreadsheet: Your Best People Are Watching
Here is the part that does not show up in any turnover-cost model, and the reason this is a mistake article and not a cost calculator.
Your A-players know exactly who is underperforming. They knew before you admitted it to yourself. And every week you protect that person, your best people learn something about you: that the standard is negotiable, that effort and outcome are optional, that the bar is wherever the weakest tolerated performer sets it. One protected underperformer does not stay contained to one seat. It recalibrates the entire team's sense of what "acceptable" means — downward.
Gallup's workplace research has consistently found that a large majority of employees are not engaged at work, and that the single biggest swing factor on a team is the manager and the standards the manager enforces. When a high performer watches a low performer face no consequence, the rational response is not to work harder — it is to quietly lower their own output to the new visible ceiling. You are not keeping one mediocre employee. You are teaching your best ones to be mediocre.
This is why "they're not hurting anyone, they're just not great" is the most dangerous sentence an operator can say. They are hurting someone. They are hurting the three people who are great, who are now wondering why they bother. A $2.8M GMV apparel operator lost their best merchandiser — not to a competitor's offer, but in an exit conversation where the merchandiser said plainly: "I was tired of carrying [name] and watching nothing happen." The mis-hire cost the founder the good hire. That is the compounding nobody prices in.
If you have built your team thoughtfully — and the sequencing in the eCommerce team structure guide assumes you have — then every role is load-bearing. A weak node in a small team is not a rounding error. It is structural.
The Regret Test: The Question That Tells You What You Already Know
Operators overcomplicate the fire decision because they conflate two separate questions. Question one is have I been fair? Question two is is this working? Those are not the same, and answering the first does not answer the second.
Here is the test that cuts through it. Imagine this person walks into your office tomorrow and resigns. Do you feel relief or loss?
If your honest, gut-level first reaction is relief — a small internal exhale, the immediate thought of "okay, here's how I'll redistribute the work" — you already have your answer. You have had it for a while. The relief is your accumulated judgment surfacing before your rationalizations can re-engage. If the reaction is genuine loss, the protective scramble of "wait, no, how would we replace them" — that is a person worth investing in, and the conversation you need is about coaching, not exit.
The regret test works because it bypasses the two cognitive traps that keep operators stuck. The first is sunk cost: you have invested months of training and you do not want to "waste" it. But the months are already gone whether you keep or release — the only question is whether you keep spending. The second is conflict avoidance, which is the real reason most firings are late. Operators are not confused about the decision. They are avoiding the conversation. Naming that honestly is the unlock: you are not waiting for more data. You are waiting to feel better about a decision you have already made.
Run the test cold, in writing, away from the person. If the answer is relief, your job is no longer to decide. It is to execute fairly.
Why Operators Wait Too Long — and the Three Stories That Keep You Stuck
Late firing is so universal that it deserves to be named as a pattern, not a personal failing. Research summarized by Harvard Business Review on managerial decision-making points repeatedly at the same culprits: optimism bias (the belief that the next quarter will be different), and the disproportionate psychological weight of the conversation versus the ongoing cost of avoiding it. Knowing the pattern is how you catch yourself inside it.
Three stories keep operators stuck, and each has a counter:
"They might turn it around." Sometimes true — which is exactly what the improvement window (below) is for. But "might turn it around" is only honest if you have given clear, measurable, time-boxed criteria and watched them lapse. If you have been saying "they might turn it around" for four months with no defined criteria, that is not patience. That is avoidance wearing patience as a costume.
"I don't have time to hire a replacement." This is the first-hire-vs-outsource decision in disguise, and it deserves a real answer rather than panic. The empty seat is often cheaper than the filled-wrong one, because the empty seat does not drag your A-players or absorb your management time. Sometimes the right move post-exit is not a like-for-like rehire at all — it is to revisit when to hire versus automate and ask whether the role should exist in its current shape.
"It's not that bad." This one you test with the regret question. If a resignation tomorrow would bring relief, it is, in fact, that bad — you have just normalized it. And the founder who is underpaying themselves to keep a below-bar seat funded should read the founder salary reality and ask who is actually subsidizing whom.
The Clean Exit: Document, One Window, Decisive Action
Knowing you should act is not permission to act sloppily. The regret test tells you the decision; this is the process. A clean exit protects the person's dignity, your remaining team's trust, and your legal position.
Document before you decide it's terminal. Write down the specific gaps — missed targets, quality failures, behavioral patterns — with dates and examples, not adjectives. "Underperforming" is a feeling. "Shipped 3 of 8 committed product launches in Q1, two with errors requiring rework" is a fact. This record is what separates a defensible decision from an emotional one, and it is what employment guidance from the U.S. Department of Labor and standard HR practice expect you to have. Documentation is not bureaucracy. It is the difference between a decision you can stand behind and one you'll second-guess.
Give exactly one improvement window — with measurable criteria. Not three windows. One. Sit down, name the specific gaps, and define what "fixed" looks like in numbers and dates: these three metrics, at this level, within 30 days. This serves two purposes. It gives a genuinely salvageable person a fair, unambiguous shot. And it removes your own ambiguity — when the window lapses without the criteria met, the decision makes itself and the conversation avoidance has nowhere left to hide.
Act decisively when it lapses. If the criteria are not met, do not start a second window. A second window is not generosity — it is the avoidance pattern reasserting itself, and your team will read it precisely as the signal you do not want to send. Move quickly, treat the person with respect, be generous and clean on the exit terms within your means, and resist over-explaining. The kindest firing is a fast, fair one. The cruelest is the slow one where everyone — including the employee — knew for months.
What This Costs You If You Get It Wrong
Get the fire decision wrong by acting too fast and you lose a person who would have grown. That is a real cost, and the documentation-plus-one-window process exists to prevent it.
Get it wrong by acting too slow — the far more common error — and the costs stack in an order most operators never trace: first the salary, then your management time, then your team's standards, then your best person's patience, and finally your own conviction that you are running a real operation. By the time all five have compounded, the severance you were avoiding looks trivial against what the delay actually took.
The decision nobody teaches operators is not how to fire. It is when. The answer is almost always sooner than your conflict-avoidance will allow, and the regret test is how you hear that signal over the noise of your own rationalizations. Run it tonight. If the honest answer is relief, document, set one window with real criteria, and then act when it lapses. The cost of waiting is the one cost on this entire page you can stop paying tomorrow.
