Full walkthroughs — measurement design, model, results, and honest limitations — using synthetic or illustrative data. Not client work, but the same approach.
Most organizations already have a decent turnover predictor — the engagement survey — and stop there. Widening the lens to pay competitiveness, promotion stalling, and peer attrition lifted holdout AUC from 0.66 to 0.77.
Read →"High potential" drives who gets development budget and succession slots, and most companies assign it by gut feel. A worked example of defining the construct, building a rubric, and checking whether two raters actually agree.
Read →Tenure and last year's performance rating barely beat a coin flip at predicting who gets promoted. A structured behavioral rubric and a validated assessment did meaningfully better — and the write-up is honest about where it still falls short.
Read →