The People Analyst Guide to Nine Lies About Work. 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 first lie is "people care which company they work for." The book's counter is that the company is an abstraction employees rarely experience directly; what they live in is the team. The brand, the values poster, the all-hands — these are company-level signals, but the daily reality of work (do I know what's expected, do I have what I need, do I trust the people around me) is set locally. So the unit that predicts whether someone is engaged, performs, and stays is the team, not the company.
What the research actually says
This is well supported, with one honest split. The split: the company does matter for attraction — brand, reputation, and mission pull people in the door, and that's real. But for everything after the door — engagement, performance, retention — the variance is overwhelmingly local. Gallup's long engagement program and the ADP Research Institute's global studies the book draws on find that the large majority of the differences in engagement sit between teams inside the same company, not between companies. Two teams under one logo, one comp plan, one set of values can look like different organizations — because they are, where it counts. The manager and the team set the conditions; the company sets the average those local conditions vary wildly around.
The measurement implication is the part organizations get backwards. A company-level engagement number is an average over enormous within-company variance — it tells you almost nothing actionable, because the problem and the fix are both at the team. Reporting one company score is like reporting one national temperature: technically a number, operationally useless.
How you run it
Measure and act at the team. Make the team the unit of analysis, not the company or the function: team-level engagement, team-level conditions, team-level trends. Resist the rollup — the rollup is where the signal dies. And resolve "team" honestly (people work on more than one; a clean team definition is its own small problem worth solving before you measure).
The analysis you can run
A team-level engagement + segmentation analysis — survey-orchestrator with segmentation-studio —
that measures the conditions where they actually vary (the team), surfaces the spread within the
company that a single score hides, and points leadership at the specific teams that need attention rather
than a company-wide program that treats every team as the average. (Pairs with the CAMS binding-constraint
work, which is itself a team-level diagnostic.)
The AI-era turn
The annual, company-wide engagement survey is the artifact of this lie — one number, once a year. AI makes the alternative cheap: a continuous, team-level read for every team, light enough to run often, with the noise handled (the rater discipline from Lie 6 applies — a team pulse you can't trust is worse than none). The shift is from one slow company average to many live local reads.
What to do Monday
- Stop leading with the company engagement number. Report the distribution across teams — the spread is the story.
- Make the team the unit of every people metric you can; resist the rollup.
- Point interventions at the specific teams that need them, not a company-wide program.
- Fix your team definition before you measure (who's actually on which team) — it's load-bearing.