Performance Management

Performance Management

The performance-management ladder is the hardest part of being a manager. Most managers avoid it for too long, then escalate too fast. The cost of avoidance is paid by the team (who watches the manager tolerate it) and by the underperformer (who deserves clarity earlier).

This skill helps the user navigate the ladder deliberately. It is not for healthy feedback loops — that's feedback-frameworks. This is for the moment when feedback alone hasn't worked.

Zeroth: read the documentation trail

If the report is a registered stakeholder, read their ~/bettersense-work-reflections/<category>/<slug>.md before diagnosing. It matters here more than anywhere else in the bundle:

• It corrects the recency story. Managers arrive at this skill with the last three weeks vivid and the last six months blurred. Dated entries show whether this is a recent dip or a long pattern — which often changes the stage diagnosis.
• It is the contemporaneous record. Entries written near the events are the honest answer to "when did you first raise this, and what did you say?" If the file shows the user noticed for months but never delivered the feedback, surface that gently — the ladder starts at Stage 1/2 regardless of how long the private worry existed. Documentation of concern is not the same as delivered clarity.
• It feeds HR-grade documentation later. At Stage 2+, the dated observations become the specifics that formal documentation requires.

Write-back discipline: from Stage 2 onward, every formal step — the conversation, the stated bar, the deadline, the report's response — gets a dated entry in the stakeholder file the same day. Prompt this at the end of each stage's conversation. Future-you (and HR) will need the record; memory will not survive the stress.

First: diagnose where on the ladder this is

Before anything else, settle which problem the user is actually solving. The interventions are different.

Push the user to name the stage. Most managers describe a Stage 3 problem in Stage 1 language ("they just need a little more support") because Stage 3 is uncomfortable. Surface that gently.

Stage 1: Coachable gap

The report has a real gap, but they're engaged, self-aware, and will close it with the right support.

• Don't escalate. Most coachable gaps don't need formal infrastructure. Hand off to coaching-mode for the conversation.
• Define what "closed" looks like. A vague "we're working on communication" is not a goal. "In the next 6 weeks, you'll lead the design review for the migration project, including handling the disagreement that's likely to come up about scope."
• Set a check-in cadence. Weekly 15-min check-ins specifically on the gap. Not folded into normal 1:1s — the separation matters.
• Document lightly. A note in your stakeholder-reflect file is enough at this stage.

Stage 2: Persistent feedback gap

You've given the same feedback at least twice. The report seems to hear it, but the behavior isn't changing. This is the most dangerous stage to mishandle — too soft and it slides into Stage 3; too hard and you blow up a relationship that could have recovered.

The conversation:

"I want to flag something. I've raised [specific behavior] with you in [date 1] and [date 2]. I'm not seeing the change I expected, and I want to make sure we're aligned that this matters and on what 'changed' looks like. Can you walk me through how you've been thinking about it?"

Then listen. Three common discoveries:

• They didn't realize it was that important. Re-state the bar and the consequence of not meeting it. Get verbal commitment to a specific behavior change with a deadline.
• They tried but the strategy didn't work. Different problem — they need a different approach, not more feedback. Pair with coaching-mode.
• They disagree it's a problem. Now you have a calibration disagreement, which is its own conversation. Surface the evidence; ask what would change their mind.

Documentation at Stage 2 becomes important. Write a follow-up email summarizing the conversation, the agreed-on changes, and the timeline. Cc nobody, but the email exists. "Per our conversation today, we agreed that…" — this is the paper trail HR will ask about if Stage 3 happens.

Stage 3: Genuine underperformance — the PIP

A PIP (Performance Improvement Plan) has a bad reputation because most PIPs are setup-to-fail. A real PIP is a fair attempt to help someone succeed; a fake PIP is documentation theater for a decision already made.

A fair PIP has:

1. Specific behaviors or outcomes, not adjectives. Not "improve communication." Yes "In the next 60 days, send a written status update every Friday by EOD that includes blockers, decisions taken, and asks. Not late more than once."
2. A reasonable timeline. 60-90 days is standard. Less than 30 is a setup; more than 120 is a different problem.
3. A measurable bar. Both parties should be able to look at the artifact at week 8 and agree whether the bar was met.
4. Real support. Weekly 1:1s focused on the PIP. Removal of distractions. Honest answers to "what would help?" If the user can't articulate the support, the PIP is incomplete.
5. Honest stakes. Tell the report that not meeting the bar means termination. Hiding this makes the PIP feel like surveillance, not support.

The PIP conversation:

"I want to talk about your performance. Based on [specific evidence], I don't think you're performing at the level your role requires. I want to put us on a Performance Improvement Plan to give you a clear shot at closing the gap. The honest stakes: if we don't see the changes in the next 60 days, we'll have to part ways. I want you to succeed here, and I'm going to support you, but I also want to be honest with you about where we are."

This conversation should not be a surprise. If it is, you skipped Stage 2. Go back.

The PIP should be in writing. HR usually owns the template; if not, write it: behaviors/outcomes, support, check-in cadence, success criteria, consequences.

Stage 4: The brilliant jerk

High output, toxic behavior. The trap most senior managers fail at: tolerating because the output is good. The cost is the rest of the team.

• Treat the behavior the same way you'd treat output failure. If the bar is "ship and don't make the team worse," then making the team worse is failing the bar, regardless of shipping.
• Document specific behaviors and the team-level effects. "On [date], in standup, you publicly described Priya's design as 'amateur'; she stopped contributing in design discussions for the next 3 weeks." Paper trail.
• Run the cost-benefit honestly. What's the team's velocity loss from people walking on eggshells? What's the attrition risk of your strongest engineers? Often the math is brutally clear once written down.
• Have the conversation directly. "Your output is excellent. Your behavior in [specific situations] is corroding the team. I'm not going to choose between them — I need both. Here's what changes, and here's the timeline."
• Be willing to lose them. A brilliant jerk who won't change is a strong-no on retention, even if HR resists. The team you build after they leave is usually better than the team they're in.

Stage 5: Mismatch (not a performance problem)

Sometimes the issue isn't that the person is failing — it's that the role is wrong. Common patterns:

• A senior IC promoted to manager who doesn't enjoy it
• An engineer in a domain that's drifted away from their interests
• A role whose scope outgrew the person's stage

This is the kindest stage to identify. The conversation:

"I don't think this role is the right shape for you. I want to talk about whether there's a different role — here, or somewhere else — that fits better. This is not a performance conversation; you're not failing. The role is the wrong fit."

Outcomes:

• Internal role change. The user advocates for a transfer.
• External role change. Honest help with a transition timeline and a reference.
• Role redesign. Shrink the scope to what's actually working.

A mismatch handled with grace builds the user's reputation as a manager people want to work for, even on the way out.

Stage 6: Termination

If Stage 3 hasn't worked, or if the offense is disqualifying (gross misconduct, harassment, illegal behavior), termination is the right move.

• Loop in HR and legal early. Always. Severance, last day, equity vesting, return of equipment, access revocation — they'll quarterback it.
• The conversation is short. "This is the conversation we discussed at the PIP kickoff. We did not see the changes we agreed on. Today is your last day." Don't relitigate the PIP; the time for that was the PIP.
• Be human. Acknowledge the difficulty. Thank them for what they contributed. Don't pretend it's anything other than what it is.
• Communicate to the team that day or the next. Tone: "X is no longer with the company. We wish them well. Here's how we're covering the work in the short term." No details. The team will infer; that's fine.
• Audit yourself afterward. Could this have been caught earlier? Was the loop selecting for this? Surface the lessons; don't bury them.

Operating principles

• Avoidance is its own decision, with its own cost. A team watches its manager tolerate underperformance and updates their model of the bar. The cost compounds.
• Documentation is kindness. It protects the report (clarity on what's expected) and the manager (when HR asks "what have you done?"). Light at Stage 1, heavier through Stage 3.
• Speed of escalation matters less than clarity at each stage. Stage 1 with clear stakes can resolve in 4 weeks. Stage 1 left fuzzy can drag for a year and become Stage 3.
• The reports you don't manage out are your message to the team. Tolerating Stage 3 tells your A-players that the standard is lower than they thought.
• Termination is rarely the wrong call once you've done a fair PIP. Almost no manager later wishes they'd held on longer. Many wish they'd moved faster.
• Calibrate with peers and your manager. Performance bars drift in isolation. Check your read against other senior ICs / managers.

Anti-patterns to flag

• Stage 3 conversation as a Stage 1 conversation. "Just give them more feedback" — when you've already given the feedback five times — is avoidance, not patience.
• The PIP that's actually a fire date. If the user has already decided to terminate, a fake PIP is cruel and legally risky. Don't do it. Go directly to a separation conversation with HR.
• Withholding information from the report. They deserve to know they're on a PIP. They deserve to know what "succeeded" means. Hiding either is unfair.
• Tolerating the brilliant jerk because they ship. Calculate the cost in attrition + chilled team contributions. Run it past your manager.
• Talking to the report's peers about the situation. Beyond a tight HR/manager loop, this is gossip. It corrodes trust if it leaks.
• Letting compensation conversations replace performance conversations. A small bonus or a flat raise is not feedback. The report should know where they stand on performance separately from comp.

Composition with other skills

• feedback-frameworks — for the actual delivery of feedback at any stage. Use COIN structure for the conversation; SOLID quality check on the draft.
• coaching-mode — for Stage 1 coachable gaps. Stay in question-mode rather than prescription-mode.
• one-on-one-prep — the dedicated PIP check-in is a separate meeting from the regular 1:1; design both.
• decision-log — log the PIP decision, the termination decision, and the calibration notes. These will matter.
• stakeholder-reflect — the running entries on a struggling report are exactly the documentation HR will ask for. Keep them current.
• leadership-os (Mirror mode) — for delivering the hard feedback with both candor and care.