Performance and Feedback for Engineering Managers (2026 Field Guide)

Hogan's feedback equation: the most-shared single artifact

Lara Hogan's feedback equation (Resilient Management, A Book Apart, 2019, chapter 3) is the most-shared single artifact in modern engineering management. The structure:

Observation + Impact + Question or Request

The mechanics:

1. Observation. A specific, factual statement about an action or pattern. 'In yesterday's design review, you cut off Lakshmi twice when she was making a point.' Not 'you've been dismissive lately' (interpretation, not observation). Not 'people feel you don't listen' (other-people's-attribution, not observation).
2. Impact. A specific statement of the consequence. 'It meant her concern about the API contract didn't get heard, and it modeled an interruption-tolerant culture for the junior engineers in the room.' Not 'it was rude' (judgment, not impact). Not 'people are bothered' (vague impact).
3. Question or Request. Either a question to surface the other person's perspective ('what was going on for you in that meeting?') or a specific request for a future change ('I want you to wait until people finish their points before responding'). The closing turns the conversation from broadcast to dialogue.

The equation works in three contexts: the regular 1:1, the in-the-moment correction, and the formal performance review. Hogan's chapter walks through each. The discipline is enforced by writing the equation on a notecard before the conversation and refusing to deliver feedback that doesn't fit the structure.

The complementary BICEPS framework (also Hogan, Resilient Management) — Belonging / Improvement / Choice / Equality / Predictability / Significance — is the diagnostic framework for understanding what the report is actually asking for in a 1:1 or performance conversation. A report saying 'I don't feel heard' is often asking for Belonging or Equality; a report saying 'I'm bored' is often asking for Improvement or Choice. Reading the BICEPS layer underneath the surface complaint produces materially better feedback conversations.

Larson's 'Useful PiP' framing

Will Larson's 'Useful PiP' essay (lethain.com, archived in An Elegant Puzzle and the lethain archive) reframes the performance improvement plan as a structured opportunity, not as a firing tool. The framing:

1. The PIP is a final-step structured opportunity to either improve or part ways with dignity. It is not a hidden firing process. The mechanic is: explicit, written, time-bound (typically 60–90 days), with named behavioral targets and a named coach.
2. The PIP is preceded by months of feedback. A PIP that arrives as a surprise to the report indicates the manager has failed to provide feedback continuously. Hogan's feedback equation applied 6–12 months earlier is the prevention.
3. Either outcome is acceptable. The report improves and stays. The report does not improve and exits with severance and a clean reference. Larson's framing: a PIP that 'fails' is not a manager failure; the manager failure is the avoidance of the conversation in the first place.
4. The team notices the action. Strong ICs on the team have been frustrated for months that the underperformer was unaddressed. The PIP signals that the manager is engaging. Charity Majors's writing on this dynamic: the team's strong ICs were the ones at risk of leaving while the manager avoided the conversation.
5. Failure mode 1: delaying. The most-cited EM failure mode. The manager waits one more quarter. The team's strong ICs interview elsewhere. The underperformer's situation worsens. By the time the PIP arrives, two of the strong ICs have already accepted offers.
6. Failure mode 2: weaponizing. The manager uses PIP to push out a report they don't like, with vague behavioral targets and a 30-day timeline that's structurally unwinnable. This produces legal risk, team-trust loss, and an HR partnership that becomes adversarial. The PIP must be honest — either improve or part ways — not theatrical.
7. Failure mode 3: wandering. The PIP is structured but the manager doesn't follow through with the weekly cadence. The report doesn't know whether they're tracking against the targets. The exit conversation, if it happens, is harder than it needed to be.

Larson's most-cited line on this: 'a properly-scoped PIP either restores someone or lets them exit with dignity. The failure mode is the manager who delays — the team notices and quits around you.'

Grove's task-relevant maturity in feedback

Andy Grove's task-relevant maturity (TRM) framework (High Output Management chapter 3, 1983) is the load-bearing framework for calibrating feedback to the task and the report. The mechanics:

1. The principle. Management style should match the task-relevant maturity of the report. A report who is high-TRM for a task (has done it before, knows the patterns, can make decisions) needs delegation and air-cover. A report who is low-TRM for a task (new to it, unfamiliar pattern, navigating ambiguity) needs structure and explicit direction.
2. The same report has different TRM on different tasks. A senior IC who has shipped 5 distributed-systems projects is high-TRM on the next one — delegate. The same senior IC running their first cross-functional partnership with a new product team is low-TRM — provide structure on the partnership rhythm and the difficult-conversation patterns.
3. TRM is about evidence, not seniority. A junior IC who has shipped two services and a senior IC who has shipped one of the same kind have similar TRM on that task. Treating them differently because of seniority alone is the failure mode the framework names.
4. The feedback shape changes with TRM. High-TRM report: 'how is the X project going?' (open-ended, leaves them to surface what's relevant). Low-TRM report: 'walk me through the data-pipeline architecture you're proposing — I want to see how you're thinking about it.' (structured, surfaces what they may not know they need feedback on). Same person, different task, different conversation shape.
5. Failure mode 1: under-delegating. The manager who treats every senior IC as low-TRM and provides structure where it's not needed. The senior IC experiences this as micromanagement.
6. Failure mode 2: over-delegating. The manager who treats every report as high-TRM. The low-TRM-for-this-task report flounders, doesn't know what good looks like, and becomes a performance issue 6 months later.
7. Failure mode 3: confusing TRM with respect. Some managers feel 'providing structure' is disrespectful to senior reports. Grove's framework rejects this directly: the structure is calibrated to the task, not to the person's status. A senior IC navigating a new domain genuinely benefits from structure, and treating them as if they shouldn't is a failure of management.

The 1:1 as the load-bearing artifact

Across every tier of engineering management, the 1:1 is the load-bearing artifact for performance and feedback. Fournier's chapter 5 (The Manager's Path), Lopp's Managing Humans (multiple chapters on 1:1 mechanics), and Hogan's Resilient Management chapter 1 all treat the 1:1 as the primary mechanic. The canonical mechanics:

1. The agenda is theirs, not yours. Lopp's framing: 'this is your time, not mine.' The report drives. The manager listens, asks questions, provides feedback when invited or when the feedback equation applies. The 1:1 that becomes a status-update channel is a misuse of the artifact.
2. Cadence. Weekly 30-min for a direct report at line-manager level. Some senior reports prefer bi-weekly 60-min. Some new hires need every-other-day for the first 2–3 weeks. The cadence should match what produces value, not what fits a calendar default.
3. Career conversation overlay. Once a month or once a quarter, a longer 1:1 dedicated to career — Hogan's BICEPS frame, Larson's writing on career-conversation mechanics, Fournier's chapter on career growth. The career conversation is structurally different from the operational 1:1 and benefits from explicit framing.
4. Documentation. Most experienced EMs keep private 1:1 notes — Lopp's writing on this is direct. The notes are for the manager's memory and for performance-cycle documentation later. They are not shared with the report. They are private but should be honest enough that the report's perf-cycle documentation can be back-derived from them.
5. The bidirectional feedback ask. Hogan and Lopp both: ask the report for feedback on you regularly. 'Is there one thing I'm doing or not doing that I could change to make our work together better?' The first time you ask, the answer will be 'nothing.' The fifth time, the answer will be useful. Discipline.
6. The skip-level signal channel (for managers of managers). Larson and Fournier both: skip-level 1:1s are not 1:1s in the same shape — they are calibration channels, lower cadence (every 6–12 months for most ICs), with privacy structure that prevents the manager-of-managers from undermining the line-manager. Larson's 'The perfect skip-level meeting' essay is the canonical operational reference.