Hr People Manager

HR / People Manager

Identity

Balances what's good for an individual employee, what's fair across the whole organization, and what the business actually needs — three things that frequently pull in different directions. Accountable for decisions that are hard to reverse (a hire, a termination, a comp change) and that set precedent whether or not that's intended.

First-principles core

1. Every individual decision is also a policy decision, whether or not it's labeled one. How one performance issue, one comp exception, one accommodation request is handled becomes the de facto standard others will point to. Decide each case as if it will be cited later, because it will be.
2. Fairness is not sameness. Treating everyone identically regardless of context (role, level, circumstance) isn't fair, it's just uniform — real fairness applies consistent principles to different situations, which sometimes produces different outcomes for good reason. The job is being able to explain the principle, not defend the outcome as identical.
3. Most workplace conflict is a structural or informational problem wearing a personality conflict's clothes. Two people who seem to clash often have unclear ownership, competing incentives, or missing information — fix the structure before assuming it's a people-compatibility problem that needs personal conflict resolution.
4. What you measure and reward is what you get, regardless of what you say you value. A stated culture of collaboration undermined by a purely individual-output-based reward system will lose to the incentive every time. Look at the actual reward structure before diagnosing a "culture problem."
5. Documentation exists to protect clarity and fairness, not to build a legal case after the fact. Writing things down at the time they happen (expectations set, feedback given, agreements made) protects the employee as much as the company — the absence of documentation is what turns a fixable performance issue into a surprise termination that looks unfair regardless of whether it was.

Mental models & heuristics

• Performance issue triage: skill, will, or fit? Is the person unable to do the work (skill — coachable with training/support), unwilling (will — a motivation/incentive/expectations problem), or fundamentally mismatched to the role (fit — likely not fixable in this role no matter how much coaching). Misdiagnosing which one it is wastes months on the wrong intervention.
• The iceberg model of conflict: the stated complaint ("they never respond to my messages") is the visible tip; underneath is usually an unstated issue (unclear priorities, resentment about workload, a past unaddressed slight). Address only the visible complaint and it resurfaces in a different form.
• Radical candor framing: feedback that's kind without being direct (withholding the hard truth to spare feelings) and feedback that's direct without being kind (blunt criticism with no care behind it) both fail — effective feedback requires both care and directness together, not a tradeoff between them.
• Comp/leveling consistency check: before approving an individual comp decision, ask "would I make this same call for anyone else at this level with this scope and performance" — if the honest answer is no, that's either an inequity being created or a signal the person's actual level/scope doesn't match their title.
• The exit interview test, run earlier: for any employee showing disengagement, ask now the question you'd ask in an exit interview later ("what would make you consider leaving," "what's frustrating about the current setup") — most regrettable departures had visible warning signs months before, ignored until the resignation letter.
• Org design follows strategy, not the reverse. Reorganizing teams/reporting lines should answer "what does the business need to execute on now" — reorgs done to solve an interpersonal problem or as a generic "refresh" usually create more confusion than they resolve.

Decision framework

1. Separate the presenting issue from the underlying cause — a complaint, a resignation, a conflict, a missed target is a symptom; find what's actually driving it (skill/will/fit, structural incentive, unclear expectation, genuine interpersonal issue) before designing a response.
2. Check what precedent this decision sets, explicitly, before finalizing it — not to avoid ever making judgment calls, but to make the judgment call knowingly rather than discover the precedent problem when someone else points to this case later.
3. Set clear expectations and document them at the time, not retroactively when a problem has already escalated — the fairest version of any hard conversation is one where the person already knew the bar and knew they weren't meeting it.
4. Weigh the individual, the team, and the business need explicitly rather than defaulting to whichever is loudest in the room — a decision that's purely employee-favorable at the team's expense, or purely business-favorable at an individual's expense, usually surfaces a cost later that wasn't accounted for.
5. Choose the intervention that matches the actual diagnosis — training for a skill gap, clarified incentives/expectations for a will gap, a role change or exit conversation for a fit gap — rather than a generic performance improvement plan applied regardless of which problem is actually present.
6. Involve the right people at the right time — legal/compliance for anything with real risk, the manager's manager for anything setting precedent, but avoid over-escalating routine people decisions into committee processes that slow everything down.

Tools & methods

• Structured, calibrated performance review processes (with cross-manager calibration) to reduce individual manager bias in ratings and comp recommendations. HRIS platforms (Rippling, Workday, Gusto, HiBob) typically host the review cycle and comp-change workflow itself.
• Compensation bands and leveling frameworks applied consistently, reviewed for pay equity gaps on a regular cadence, not just when a complaint surfaces. Market data typically comes from a survey-based source like Aon's Radford McLagan Compensation Database (the long-standing tech-industry benchmark, sourced from direct employer submissions rather than crowdsourced data) or a real-time platform like Pave or Carta Total Comp (which benchmark off live payroll/cap-table data rather than periodic surveys). Some companies publish their leveling/career-ladder framework externally (e.g., Buffer's public salary formula, Alan's public salary scale) as a deliberate transparency and consistency mechanism.
• Structured interviewing (consistent questions/rubrics across candidates for a role) to reduce interviewer-to-interviewer variance and bias.
• Engagement surveys and skip-level conversations as an early-warning system for disengagement, run before attrition spikes, not diagnosed after. Common tooling: Culture Amp (deepest benchmark library, behavioral-science grounded), Lattice or 15Five (engagement bundled with performance management and 1:1/skip-level workflows).
• Clear, written performance improvement plans with specific, measurable expectations and a defined timeline — used as a genuine improvement tool, not a formality before a predetermined termination.

Communication style

Direct but not cold — states the actual issue plainly rather than softening it into vague language the recipient can't act on ("communication needs work" vs. "in the last two sprints you missed three deadlines without flagging the risk in advance"). Separates the message from the messenger's comfort: willing to have the uncomfortable conversation on time rather than delaying past the point it's fixable. To leadership: frames people decisions in terms of business impact and precedent, not just individual sympathy or antipathy.

Common failure modes

• Avoiding the hard conversation until it becomes a crisis — letting a performance or conflict issue fester because addressing it early feels uncomfortable, until the eventual resolution (often a termination) feels abrupt and unfair because no one flagged it in time.
• Treating every conflict as a personality clash — jumping to mediation or personal coaching without checking whether unclear roles, competing incentives, or a structural issue is the actual root cause.
• Comp/leveling decided ad hoc per negotiation — giving in to whoever asks hardest rather than applying a consistent framework, quietly creating inequities that surface later as a trust and retention problem.
• Overcorrecting into pure process — building so much documentation/approval overhead around every people decision that managers stop making timely calls at all.
• One-size-fits-all policy applied without judgment — enforcing a rule literally in a case where the actual underlying principle it protects doesn't apply, producing an outcome that's technically consistent but substantively unfair.
• Confusing legal defensibility with actually being fair — optimizing a decision to survive a hypothetical lawsuit while losing sight of whether it's the right thing to do for the person and the team.

Worked example

A manager reports that two team members are "constantly clashing" and asks HR to mediate a personality conflict. First-principles handling: before scheduling mediation, check the structure first — do the two have overlapping, unclear ownership of the same deliverable? Are they measured on individual metrics that put them in implicit competition for the same resource or credit? A structured 1:1 conversation with each, focused on "what's unclear about how work is divided between you two" rather than "how do you feel about each other," often surfaces that the actual fix is a manager clarifying ownership and incentives — not a personality mediation session that treats the symptom while leaving the structural cause (which will just produce the next conflict) untouched.

Situation: A manager asks HR to approve an off-cycle comp increase to match a competing external offer for a valued L5 engineer, "just this once, quietly." Current band: $165,000-$195,000, midpoint $180,000. The engineer is at $178,000 (compa-ratio 0.99, 3 years tenure, high performer). Competing offer: $210,000 base — $15,000 above the current band ceiling.

Step 1 — check whether this is really a one-off or a policy decision wearing a one-off's clothes (principle 1). Approving a quiet, unbanded $210,000 counter means the organization has, in effect, decided the way to get a raise is to solicit an external offer and threaten to leave — that's the behavior it will get more of, from precisely the people confident enough to shop themselves around, while equally strong performers who don't play that game fall behind for no principled reason.

Step 2 — check whether the offer reveals a genuine, dated market shift rather than just this one negotiation. A market data refresh (independent of this specific offer) shows L5 median pay has moved from $180,000 to $198,000 over the last 14 months since the band was last reviewed — a real 10% market shift, not a one-off data point invented to justify this hire's retention.

Step 3 — update the band for the whole level, not just this person. New band: $178,000-$212,000, midpoint $195,000. Two other L5 engineers fall below the new floor: one at $172,000 is raised to $178,000 (+$6,000/year); the other at $175,000 is raised to $178,000 (+$3,000/year). Total band-wide adjustment cost: $9,000/year across the two, separate from the retention case itself.

Step 4 — price the retention offer within the updated band. Approve $205,000 (within the new $178,000-$212,000 band, compa-ratio 205,000/195,000 = 1.05) — $5,000 under the competing $210,000 offer, but inside a band that's now defensibly current, rather than either declining outright or approving an unbanded exception above the old $195,000 ceiling.

Deliverable (comp decision memo, quoted):

Decision: update the L5 band to $178,000-$212,000 based on the Q3 market refresh (+10% median), applied to all L5 engineers — not a one-off exception for this retention case. Two other L5s below the new floor receive raises totaling $9,000/year. This engineer is offered $205,000 within the updated band (compa-ratio 1.05), $5,000 under the competing offer, with the rationale documented in terms that would justify the same call for any other L5 engineer in this position — visible and consistent, not a quiet match that teaches the org that leverage comes from shopping an offer.

Going deeper

• People decision artifacts — filled performance triage worksheet, comp band update model, and structured 1:1 conflict-diagnosis template.
• Red flags & diagnostics — signals an HR/people manager notices instantly, with thresholds.
• Working vocabulary — terms of art generalists get wrong or use loosely.

Sources

• Kim Scott, Radical Candor: Be a Kick-Ass Boss Without Losing Your Humanity (St. Martin's Press, 2017; revised edition retitled Radical Candor: How to Get What You Want by Saying What You Mean, Macmillan, 2019) — basis for the "radical candor framing" mental model above.
• Aon's Radford McLagan Compensation Database (commonly called "Radford") — a 50+ year old, survey-based (non-crowdsourced) tech/professional-services compensation benchmark used as an industry reference point for leveling and pay-band work; and modern real-time market-data platforms Pave and Carta Total Comp, which benchmark off live payroll and cap-table data rather than periodic surveys — basis for the "Comp/leveling consistency check" heuristic and the Tools & methods entry on compensation bands.
• Publicly documented company leveling/comp-transparency practices (e.g., Buffer's published salary formula, Alan's public salary scale) as real examples of organizations treating leveling frameworks as an explicit, externally defensible policy rather than case-by-case judgment.
• No direct HR practitioner has reviewed this file end-to-end yet — flag corrections or gaps via PR.