The Relative Improvement Contract

Definition

An alternative to the fixed performance contract in which nobody commits in advance to a specific number. Instead, teams are evaluated with hindsight against a range of relative indicators — growth versus last year, performance versus competitors, position in an internal league table — decided by a peer-review panel after the period has ended. Because there is no fixed number to hit, the authors argue, "there is little point in gaming."

In the Book

Chapter 2's Table 2-1 sets fixed and relative contracts side by side, term for term: where the fixed contract says "your target is fixed at $x million," the relative one says "we trust you to maximize your profit potential... and remain in the top quartile of your peer group." Chapter 4 details how this plays out in practice at Handelsbanken, which uses one groupwide profit-sharing scheme (Oktogonen) rather than individual bonuses, relying on monthly published league tables of branches and regions and peer pressure rather than fear of missing a number. Groupe Bull under Jean-Marie Descarpentries used a weighted formula — growth versus prior year, growth versus competition, profit versus prior year and competition, debt, and quality — scored by an executive committee only after the year closed, so managers could not know in advance exactly what payout a given outcome would produce. The chapter likens this to a car race: each driver knows what preparation should improve their performance, but only with hindsight do they know how well they actually did relative to everyone else facing the same unpredictable conditions.

Why It Matters

Separating "what counts as good" from "a number agreed before the fact" removes the single lever that makes gaming rational — there is no fixed threshold to manipulate toward. This generalizes past corporate bonuses to any evaluation system: grading on a curve, promotion by stack-ranking against peers, or scoring an athlete by finishing position rather than a pre-declared time. The tradeoff the book is candid about is that relative evaluation trades a false certainty (a defined number) for an acknowledged subjectivity (a panel's judgment), which only works if the panel is trusted to be fair.