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 seventh lie is "people have potential." The book's objection is to potential as a fixed, portable property — a HiPo stamp you carry that says you'll succeed at bigger things — and to the whole apparatus of potential ratings and nine-box grids built on it. What people actually have, it argues, is momentum: a current trajectory, a direction and rate of travel that you can see and help, rather than a latent score you possess.
What the research actually says
The critique of "potential" is sound; the trick is keeping the development half. Sound: potential ratings are low-validity and bias-laden. "Potential" is a fuzzy construct, raters disagree on what it means (the Lie 6 rater problem in a particularly abstract form), HiPo designations predict later success weakly, and they import the rater's similarity and halo biases under a neutral-sounding label. Past performance and a credible trajectory predict future performance better than a potential tag does.
The half to keep: this is not "people don't grow." They do — that's the whole point of momentum; growth is real and is exactly what a trajectory captures. The error isn't believing in development; it's freezing a person into a static label ("high potential," "solid," "limited") that then becomes a self-fulfilling allocation of opportunity. The honest version: stop scoring a trait called potential; measure momentum — recent performance and its slope — and remember a trajectory is a measurement to update, not a verdict to assign.
How you run it
Replace the potential rating with a trajectory read. Measure recent performance and its direction over time (is it rising, flat, falling, and how fast), treat it as a quantity you re-estimate each period, and watch for the bias the old label hid — who gets the stretch assignments that create the next data point. And reliability-check it, because a trajectory built on the same noisy ratings inherits the same noise.
The analysis you can run
A performance-trajectory / distribution analysis — calculus — that models performance over time
(level and slope), surfaces real momentum against the population distribution, and replaces the static
potential axis with a measured, updateable trajectory. Pair it with the rater-reliability work (Lie 6):
trajectory is only as trustworthy as the ratings underneath it, so estimate that first. (Distribution-
aware, so it also speaks to Lie 8's "two tails.")
The AI-era turn
"AI HiPo prediction" is arriving and it is this lie's most dangerous form: a model trained on who was promoted will confidently launder the historical bias into a forward-looking "potential" score with a clean number and no error band. The discipline is the same — measure momentum against real outcomes, not a learned impression of who looks like past winners, and reliability-test the model like any rater before a career rides on it.
What to do Monday
- Drop "potential" as a trait axis (the nine-box's vertical) from promotion and succession calls.
- Replace it with measured momentum — recent performance and its slope — re-estimated each cycle.
- Audit who gets the stretch assignments that generate the next data point; that's where the old label's bias hides.
- Before trusting any AI "potential" score, check what it's really predicting and reliability-test it.