Measure What Matters: How Google, Bono, and the Gates Foundation Rock the World with OKRs

by John Doerr

TRAP #5: Low Value Objectives (aka the “Who cares?” OKR). OKRs must promise clear business value—otherwise, there’s no reason to expend resources doing them. Low Value Objectives (LVOs) are those for which, even if the Objective is completed with a 1.0, no one will notice or care. A classic (and seductive) LVO example: “Increase task CPU utilization by 3 percent.” This objective by itself does not help users or Google directly. However, the (presumably related) goal, “Decrease quantity of cores required to serve peak queries by 3 percent with no change to quality/latency/ . . . and return resulting excess cores to the free pool” has clear economic value. That’s a superior objective. Here is a litmus test: Could the OKR get a 1.0 under reasonable circumstances without providing direct end-user or economic benefit? If so, then reword the OKR to focus on the tangible benefit. A classic example: “Launch X,” with no criteria for success. Better: “Double fleet-wide Y by launching X to 90+ percent of borg cells.” TRAP #6: Insufficient KRs for committed Os. OKRs are divided into the desired outcome (the objective) and the measurable steps required to achieve that outcome (the key results). It is critical that KRs are written such that scoring 1.0 on all key results generates a 1.0 score for the objective. A common error is writing key results that are necessary but not sufficient to collectively complete the objective. The error is tempting because it allows a team to avoid the di...