The People Analyst Guide to First, Break All the Rules. Format: what the book argues → what the research actually says → how you run it → the analysis you can run → the AI-era turn → what to do Monday. No reproduction of the book's text. The substance is ours; research anchors verified on read.
What the book argues
The book's through-line — the source of its famous line — is that the manager matters most: people join companies but leave managers, and the immediate manager, not the company's policies or perks, is the strongest lever on whether a team is engaged, productive, and stays. The four keys are really manager behaviors; the unit of people performance is the team, and the manager sets the team's conditions.
What the research actually says
The strong version is well supported; the slogan is oversimplified, and a people analyst should say both. Supported: the manager explains a large, replicated share of the variance in team engagement and retention (the local-experience finding — the same one behind Nine Lies Lie 1), and managerial quality predicts unit outcomes. The conditions a team experiences are set locally, and the manager is the largest single local input.
Oversimplified: "people leave managers, not companies" is catchier than the data. Turnover is multi-causal — pay, opportunity, life events, and company factors all matter, and the manager's share, while real and large, isn't the whole story. So the honest version: the manager is the highest-leverage controllable condition, which is exactly why it's worth measuring well — not because it's the only thing, but because it's the biggest lever you can actually pull. And measuring "manager quality" runs straight into the rater problem (Lie 6) and the leadership-construct problem (Nine Lies Lie 9): measure it against outcomes, not a competency survey.
How you run it
Measure managers against what their teams actually experience and produce — team engagement conditions, retention of good people, the team's binding constraint — not a 360 personality profile. Grade against reality (the Measuring Leadership discipline), reliability-check it, and treat it as developmental (manager skill is learnable).
The analysis you can run
A manager-effectiveness analysis — manager-effectiveness with team-level engagement
(survey-orchestrator) — that scores managers on the conditions and outcomes of their teams, graded
against reality rather than perception. It's the operational form of "the manager matters": find which
managers are actually moving their teams' conditions, and develop the rest. (Braids the Measuring
Leadership battery + Nine Lies Lies 1, 6, 9.)
The AI-era turn
AI manager-assessment and coaching is arriving fast. Pointed at a "leadership competency" read it re-imports the construct + rater problems; pointed at the manager's team outcomes (did engagement conditions improve, did good people stay, did the binding constraint move) it's leverage — and AI coaching can scale the developmental half if it's aimed at measured behaviors, not vibes. Same discipline: measure against reality, reliability-test, develop.
What to do Monday
- Grade managers on their team's conditions and outcomes, not a 360 personality profile.
- Treat "people leave managers" as directionally true, not literally — manager is the biggest controllable lever, not the only cause.
- Reliability-check manager measures (Lie 6) and grade against reality (Nine Lies Lie 9).
- Make it developmental — manager skill is learnable; the measure is a starting point, not a verdict.