Português

Manager Coaching and Feedback That Actually Changes Behavior

Most "manager coaching" sessions are venting dressed up as development. The manager talks for forty minutes about how hard their week was, you nod, you ask a thoughtful question or two, they leave the room feeling lighter, and on Monday the team Slack channel still looks exactly the same. Same passive-aggressive thread, same one engineer doing all the project updates, same skip-level meeting where someone hints that Diana is "fine, I guess."

That isn't coaching. That's a therapy session you weren't trained to run.

The test is brutally simple. After every coaching session, write down one specific behavior the manager will do differently. If you can't name it, the session didn't happen. If you can name it but nothing changed in 30 days, you ran a feel-good ritual. Coaching is measured in behavior delta, not session count and not how the manager felt walking out.

This playbook is the method I use with line managers and 2nd-line managers at B2B SaaS companies. It's the part of the org where one weak manager can quietly cost you four engineers and a quarter of velocity before anyone notices.

The 1:1 Coaching Framework: SBI + Ask + Commit

Most HRBPs know SBI (Situation, Behavior, Impact). It's a great feedback frame and a terrible coaching frame on its own, because it stops where the work actually starts. SBI gets the manager to acknowledge a moment. It does not get them to change anything.

The full frame I use in every coaching 1:1:

S — Situation. Anchor to a specific recent moment. Not "in your team meetings" but "in last Thursday's sprint review at 2pm."

B — Behavior. What you observed, not what you concluded. "You interrupted Priya twice in the first ten minutes," not "you don't make space for women on the team."

I — Impact. What happened next. Engineering velocity, retention risk, a peer's read, a customer-facing miss. Concrete consequence, in their language.

A — Ask. This is the part most HRBPs skip and the entire reason behavior doesn't change. After SBI, you say plainly: "Here's specifically what I'd want you to do differently next time." Not a hint. Not a Socratic question. A direct ask. Coaching that ends in "what do you think you should do?" puts the cognitive load on the person who already showed they didn't see the problem.

C — Commit. They restate what they will do, by when, with what evidence. "Next sprint review, I will not interrupt anyone in the first ten minutes. I'll record the meeting and review it. We'll check it in two weeks."

A complete coaching exchange might run:

"In Thursday's sprint review (S), you cut off Priya twice in the first ten minutes (B). She didn't speak again for the rest of the hour, and she's the one who flagged the auth bug last sprint (I). What I want is for you to hold airtime in the first ten minutes and let people finish (A). What will you do, by when?" (C)

Manager: "I'll record next Thursday's review and not interrupt for the first ten minutes. We can check on the 22nd."

That's a coaching session. Forty seconds of structure, one behavior, one date, one piece of evidence. Everything else is wrapping.

Surfacing the Real Issue

Half the time the manager isn't avoiding the conversation about Priya being interrupted. They're avoiding the conversation that an engineer they hired isn't going to make it, or that they're underwater themselves and overcompensating by dominating meetings. The presenting problem is rarely the real one.

The unlock question I use, almost every session:

"What would you do tomorrow if I weren't asking?"

Watch what happens. They stop performing. They stop giving the answer they think you want. Nine times out of ten, what comes out next is the actual issue: "Honestly, I'd put Marcus on a PIP. I just don't want to be the bad guy."

Four follow-ups that get past performance theater once the unlock question lands:

  1. The two-week test. "If nothing changed in two weeks, what's the one outcome you'd be most worried about?" This forces specificity. Vague worry is a tell that they haven't actually thought about it.

  2. The hire-back question. "Knowing what you know now, would you hire Marcus into this role today?" If the answer is anything other than a clean yes, you have a performance conversation, not a coaching one.

  3. The avoided conversation. "What's the conversation you've been avoiding for more than 30 days?" The number matters. "More than 30 days" gives them permission to admit it without feeling like they failed yesterday.

  4. The Monday test. "If I called your skip-levels on Monday and asked them what's the one thing they wish you'd do differently, what would they say?" Most managers can answer this immediately. They already know. They just haven't said it out loud.

These aren't gotcha questions. They're permission slips. The manager already has the diagnosis. Your job is to make the room safe enough that they say it.

The Manager-NPS Signal: When to Actually Intervene

You can't coach every manager every quarter. You don't have the bandwidth and most don't need it. The question is which managers, and when.

Stop using gut feel. Start using manager-NPS.

In your engagement survey, include the question: "On a scale of 0-10, how likely are you to recommend working for your manager to a friend?" Break the score out per manager, not just org-wide. Then apply this trigger:

The 5+ / 6 rule: When a manager has 5 or more direct reports and the median score on the manager-NPS question is 6 or below, the HRBP intervenes within 2 weeks. Not at the next cycle. Not when the next 1:1 happens to be scheduled. Two weeks.

Why those numbers. Five reports gives you enough signal to filter out one disgruntled person. Six is the line where engagement starts to predict regrettable attrition within two quarters. Below 6 is a leading indicator, not a lagging one. Two weeks is the window before people who answered honestly start updating their LinkedIn.

This rule does two things. It removes your judgment from the trigger (good, because your judgment is biased toward managers you like). And it gives the VP a defensible reason for why you're putting hours into this manager and not that one. "Five reports below 6" is a sentence the VP doesn't argue with.

A second trigger worth adding: any individual report scoring their manager 0-3 is its own intervention, regardless of team size. That's a "leaving in 90 days" signal and you should treat it like one.

Feedback Delivery to Managers

How you deliver coaching to a manager matters more than the content. Three rules I will not break.

Private. Never coach a manager in front of their peers, their reports, or their boss. Even neutral observations land as public correction. If the manager needs to hear it from their VP, that's a separate meeting, and you set it up, you don't ambush.

Specific. "You're being too aggressive in standup" is unusable. "On Tuesday's standup at 9:15, when David said the API was blocked, you said 'figure it out' and moved on. That's the moment I want to talk about" is usable. The manager can replay it, see it, and change it. The vague version they will reframe by Friday into "Camellia thinks I need to be nicer."

Written follow-up within 24 hours. This is the rule HRBPs skip most often and it is the reason verbal coaching doesn't stick. A 3-bullet recap, sent in DM or email, no exceptions:

Hi [Manager],

Quick recap from today so we're aligned:

  1. What we discussed: The pattern of cutting people off in the first ten minutes of sprint review (specifically Thursday with Priya).
  2. What you committed to: Recording next Thursday's review, not interrupting in the first ten minutes, reviewing the recording yourself.
  3. When we check: Our 1:1 on the 22nd. I'll ask what you noticed.

Anything I got wrong, push back today.

— Camellia

Three bullets. No therapy language. The "anything I got wrong, push back today" line is load-bearing. It forces them to either confirm the commitment or surface the disagreement now, not three weeks later when nothing has changed and they tell you "that's not what I agreed to."

Verbal-only feedback to managers fails because managers reframe within a week. They are paid to package narrative for their teams; they will package your feedback the same way. The written recap is the artifact you both work from. Without it, you're arguing about memory.

The "Coach vs Cop" Tension

Here is the part nobody tells new HRBPs. You are coaching the manager and reporting risk on the manager to the VP. Those two roles are in tension and the manager knows it before you do.

The mistake is pretending the tension doesn't exist. The fix is naming it out loud, in session one, with this script:

"I want to be straight with you about my role here. Most of what we talk about stays in this room. That's how coaching works, and you should be able to bring me real problems. There are three things I will escalate to [VP]: a pattern of issues across two or more reports, a retention risk inside 60 days, or a values or behavior incident. If any of those three come up, I'll tell you in the room before I escalate, not after. That's the deal. Does that work for you?"

Three things happen when you say this in session one. The manager stops trying to figure out what's safe to say. You earn the right to escalate later without feeling like a snake. And you get an explicit yes, which you will need on the day you have to escalate and they accuse you of bait-and-switch.

The deal is: full confidentiality except the three triggers, no surprises, escalation is announced not ambushed. Don't promise total confidentiality. You can't keep that promise and the day you break it is the day every manager in the org learns HRBP coaching is a trap.

When to Escalate to the VP

Three triggers, no exceptions:

  1. Pattern across 2+ reports. One unhappy direct report is a relationship problem. Two with the same complaint pattern is a manager problem. Escalate.

  2. Retention risk inside 60 days. A high-performer telling you they are looking, or a regrettable resignation that came as a surprise. The VP needs 60 days, not 14, to react. Late escalation is the most common HRBP failure mode.

  3. Values or behavior incident. Anything that would be embarrassing in a deposition. Comments about a protected class, retaliation signals, a documented blowup. These never sit in your notebook. Same-day escalation, no exceptions, no "let me coach them through it first."

The escalation memo is short. One page, four sections:

Manager: Diana Chen, Eng Manager, Platform team Trigger: Pattern across 2+ reports (3 of 6 reports flagged in skip-levels over the last 30 days) Pattern observed: Specific behavior, with dates: interrupting in meetings (3 occurrences), public correction of one engineer (2 occurrences), unilateral scope changes without team input (4 sprints). Coaching to date: Two SBI sessions (Mar 14, Apr 2). One commitment met, one missed. Written recaps attached. What I'm asking from you: A direct conversation with Diana from you in the next 14 days, framed as a continuation of my coaching, not a replacement. I'll attend if useful.

That's the memo. No adjectives. No "I'm worried about." Dates, observations, what's been tried, what you need from the VP. The VP can act on this. They cannot act on "Diana seems off."

The 30/60/90 Post-Coaching Cadence

Coaching doesn't end at the session. It ends when the behavior is durable or the case is closed. Run every coaching engagement on a 30/60/90 cadence.

Day 30 — Behavior check (manager-side). In your 1:1, ask the manager: "What did you do differently in the last 30 days, and what's the evidence?" Not "do you feel like you've been working on this." Evidence. A recording, a peer's observation, a direct report's reaction in a 1:1. If they can't name evidence, you reset the commitment with sharper specifics. You do not extend the timeline indefinitely.

Day 60 — Team-signal check (their side). Do not rely on the manager's self-report. By day 60, you need outside signal. Run a skip-level with 2-3 of their reports. Pull the latest manager-NPS pulse if you have one. Look for delta, not perfection. If the team signal is moving in the right direction, keep going. If it's flat or worse, you have a decision to make at day 90.

Day 90 — Decision point. Three options, pick one:

  1. Continue coaching if the trend is positive and a clear next behavior is in scope.
  2. Escalate to the VP if the trend is flat or negative, with the memo above.
  3. Close the engagement if the behavior is durable. Send a closing recap to the manager and a one-line note to the VP. Don't let coaching engagements drag on quietly. They create false comfort.

Here's the checklist I keep on each open coaching case:

  • Day 0: SBI + Ask + Commit captured in writing, sent in 24h
  • Day 14: Mid-point check, recommit or refine
  • Day 30: Behavior check with evidence
  • Day 60: Skip-level or pulse run, delta logged
  • Day 90: Continue / escalate / close decision sent to VP

Five touchpoints over 90 days. That's a coaching engagement. Anything less is a series of well-meaning conversations that the VP will eventually ask you to account for, and you won't have an answer.

The Only Metric That Matters

Behavior change. That is the entire scoreboard. Not session count, not survey participation, not whether the manager said "this was really helpful" at the end. If, 90 days from now, the team Slack channel looks the same, you didn't coach. You hung out. The good news is the method is repeatable: SBI plus Ask plus Commit, the unlock question, the manager-NPS trigger, written recaps, the named coach-vs-cop deal, and the 30/60/90 cadence with a hard decision at day 90.

Run it on every manager who clears the trigger. Don't run it on managers who don't. And keep the file. Six months from now, when someone asks why you put 12 hours into Diana and not Marcus, the file is the answer.

Learn More