Relative KPIs and Peer Benchmarking

Definition

Rather than setting fixed absolute targets (e.g., "reduce cost 5%"), relative KPIs measure performance against peers or historical trends (e.g., cost per unit versus other plants). This creates a self-regulating system where poor performers feel urgency to improve, high performers motivate others through learning, and the measurement itself doesn't trigger gaming because the benchmark moves with business conditions.

In the Book

Borealis implemented extensive benchmarking across 30 European production plants (Chapter 3). Plants knew monthly how their operational metrics (unit cost, yield, safety) compared to comparable facilities inside and outside the company. Bogsnes describes how acceptance of low rankings was always met with denial ("our plant is different, the data is wrong") but when a plant ranked low on multiple metrics across multiple reporting cycles, "something clicked" and competitive drive awakened. The power was not in top-down pressure but in the visibility itself—no instructions came from head office, just transparent data and the implicit message: "you know what to do." High performers worked hard to maintain position; low performers worked hard to improve. This is contrasted with absolute targets, which inherently create incentives to negotiate easier baselines.

Handelsbanken (Chapter 2) uses comparable branch KPIs (return on capital, cost/income ratio, customer satisfaction) to drive both performance and knowledge-sharing. The collective bonus is tied to how Handelsbanken performs versus other banks, not to hitting fixed targets, creating mutual interest in system-wide success.

Why It Matters

Absolute targets create predictable gaming (lowballing during negotiation, "spend it or lose it" behavior). Relative metrics align incentives with actual competitive position. Because benchmarks are transparent and move automatically as peers improve, there is no target to negotiate down and no artificial floor below which you can coast. It converts measurement from a control mechanism into a learning mechanism.