A ProvenChaos Guide
Why Do I Keep Holding On to Underperformers When I Already Know What to Do?
You hold on because the delay protects you, not them. You made the call privately, often months ago. Everything since is Stealth Stalling: the CEO who has already decided stays busy around the decision instead of acting on it. Each week costs your best people’s trust and another slice of your attention.
This came up in 26 of my 572 recorded coaching sessions.
If I already know what needs to happen, why haven’t I done it?
You haven’t done it because the decision isn’t the part you’re stuck on. You made the call privately. What stalled is acting on it.
I watched a CEO spend weeks stress-testing a separation. He ran the numbers again, role-played the conversation, consulted another advisor and rebuilt the offer structure until it was airtight. Then the timeline came out. He had made the decision in December.
Every scenario he planned for was real. None of it changed the outcome he already knew was right. He wasn’t deciding anymore; he was preparing, and the preparation had become armor against a conversation he hadn’t given himself permission to have.
Each step felt responsible: one more data pull, one more advisor, one more draft of the script. That’s why it’s so hard to catch in yourself.
The test is simple. Before the next step, ask whether it could change the outcome. If it can’t, it isn’t due diligence. It’s delay with a to-do list.
Is this loyalty, guilt, or just not wanting to be the bad guy?
It’s all three, and they grow from one root: loyalty to someone’s history instead of to the seat they hold now. The guilt and the fear of being the bad guy both come from that history, and it quietly sets the bar for their current job.
It shows up in two familiar ways. You lend money to an early employee who’s struggling, and now letting them go feels like calling in a debt. You protect the year-one hire who stayed when payroll was tight, because firing them feels like betraying the company you used to be.
Both moves feel generous. They also keep a person in a role the business has outgrown, and they make the eventual conversation harder.
Honor the history out loud: thank them, credit what they built, and be generous on the way out. The seat still gets judged on what it needs today.
Why do I get pulled back in every time they improve a little?
You get pulled back because a small improvement feels like proof you were right to wait, so the clock resets and the bar drops a notch.
The slide starts with the first good week after you’ve privately decided. They hit one deadline or run one good meeting, and you think they’re turning it around. The original gap never closed. Now you’re measuring them against last month’s low point instead of what the role requires. Repeat that a few times and you’re tolerating work you would never have hired.
The bar stops sliding when the written standard and the decision date exist before the next good week shows up. A strong Tuesday doesn’t move the date. Only hitting the bar does.
How do I know it’s them and not something I failed to give them?
You know it’s them when you can show they had the tools, time and training the role required, and the results still didn’t come.
I run this as the 5Ts of Execution: tools, time, training, talent, tenacity. The first three are yours to provide. The last two are theirs to bring.
Be honest about the first three. Did they get the systems and budget the work needed? Did they have the hours, or were they covering two jobs? Did anyone train them, or did you hand over a vague role and hope?
If you gave all three, the gap is talent or tenacity. More support won’t fix either one, so stop second-guessing and act.
If you didn’t, fix what’s missing this week, then start the clock with a written bar and a decision date.
What does keeping them cost me?
Keeping an underperformer costs you twice: the person learns that waiting works, and the rest of the team watches what gets tolerated.
A construction CEO I coached walked her project manager through a $3,000 subcontractor dispute. She proposed six solutions. The project manager rejected all six, offered no counter-proposal and escalated it back up to her.
The project manager had learned that if she held out long enough, the owner would solve it. Everyone else on the team was watching that lesson get taught.
Am I hiding in strategy and values work so I don’t have to have the conversation?
Yes, if you launch a vision session, reorganization or new coaching plan after you’ve already decided about someone. That work is the conversation in disguise.
Watch the timing. The values workshop gets booked the same month their results slip again. The org chart gets redrawn so their seat moves instead of them. A fresh development plan appears for a person you stopped believing in months ago.
A CEO I coached saw his marketing lead’s output soften while leads from AI-driven channels accelerated. He responded by assigning a Q2 plan and collecting KPIs instead of saying plainly that the role was at risk, which is Stealth Stalling dressed up as management. The lead almost certainly did not understand how serious the gap was, because nobody said it.
Each move feels like leadership and buys another quarter of quiet. You already have the information you need: the results, the missed bar, your own private decision.
Before you schedule anything new, ask one question: would you still run this if that person had already left? If the answer is no, cancel it and have the conversation first.
How long is too long to give someone another chance?
Too long is any second chance that starts without a written bar and a decision date. One clear conversation, one short window, one decision: that’s the whole allowance.
Start with the conversation. Tell them plainly what the role needs, where their results fall short, and that their job depends on closing the gap. If they walk out unsure whether it was a warning, it wasn’t one.
Put the bar in writing the same day: the specific results and how you’ll measure them. Keep the window short, weeks rather than quarters. When the decision date arrives, decide.
What do I actually say when I make the call?
Say four things in order: what you expected, what happened, what you’ve decided, and what comes next. Clarity is kindness, so keep it short.
What you expected: the written bar and the date you set. What happened: the results, stated as facts, no adjectives. The decision: "We’re ending your role, effective Friday." Say it in the first two minutes. What comes next: severance, timing, how the team will hear, and who they call with questions.
Skip the softeners: "this may not be the right fit," "let’s see how next month goes," "I’m open to ideas." Each one turns your decision back into a maybe and sends them home thinking it’s still a negotiation. Don’t relitigate the evidence, and don’t defend the history.
Once you’ve said it, stop talking and let them respond.
Try this: Set a decision date and say it aloud
Pick a decision date for this person today, write the bar, and tell one person who will hold you to it before you leave the office.
Choose someone with no stake in the outcome: a peer CEO, your coach, your spouse. Send them this message as written:
"By [date], I will decide whether [role] stays. The bar is [specific result] by [date]. If it isn’t met, I’ll have the conversation that week. Please ask me about it on [date]."
Put the date on your calendar, with a reminder two days early.
For the next 30 days, run one rule. Any new data pull, advisor call or script rewrite has to pass a single test: could it change the decision? If it can’t, skip it.
When your partner asks on the date, answer with what you did, not what you planned to do.
Before you write any of it, answer the harder question: what date did you actually decide?
Use this with your team
Run this 20-minute agenda to surface the calls your leadership team has already made but hasn’t acted on. It works as a standing item in your quarterly exec-team review. Keep names out of the room. You’re looking for the pattern across the team, not building a list of people.
- Set the frame (3 minutes). Read one question aloud: where are we tolerating work we would never have hired?
- Write privately (4 minutes). Each leader writes down one seat on their team that sits below the bar, the date they first knew, and every step they’ve taken since.
- Share the gap (6 minutes). Each leader says two things out loud: how many weeks have passed since they knew, and whether any step since then could have changed the decision.
- Set the bar and the date (5 minutes). Each leader commits to writing the bar this week and states a decision date to the group.
- Close (2 minutes). One person records every date. Open the next meeting by asking for results, not plans.
Holding On to Underperformers: Common Questions
What if I’m wrong about them?
The written bar will show it. If they hit the specific results by the date, they stay, and you were wrong in the best possible way. Let the bar test your belief instead of defending it. Doubt is a reason to set the bar, not a reason to skip it.
Should I keep coaching them while the clock runs?
Yes, and coach for real, not to build a file. I call that difference Coaching versus Catching: coaching helps them reach the bar; catching only documents the miss. If you’ve stopped believing they can do it, admit you’re catching.
What if they’re a friend or a cofounder?
The bar stays the same; the care around it goes up. Have the conversation yourself and be generous on the way out. Avoiding the conversation to protect the friendship costs you both.
What do I tell the rest of the team?
Tell them the facts the departing person agreed to share, the same day, before rumors fill the silence. Skip the backstory and any criticism. Then tell them what changes in their work this week.
Should I talk to an employment lawyer first?
Yes, before the conversation, not instead of it. Legal advice can change how they exit, so it passes the "could this change anything" test. A third review of the same advice can’t, so skip it.
How these numbers were counted. This question came up in 26 of 572 recorded coaching sessions. The problems CEOs raise in Jerry’s recorded sessions (August 2024 to October 2026) are condensed by AI into short statements, with names and companies left out, and matched to this question by meaning at a conservative similarity floor checked by hand. A session counts once however often the problem came up in it, so the number is a floor, not an estimate.