Goal-Setting Theory
Goal-Setting Theory (Edwin Locke and Gary Latham, formalised 1990) is the most replicated finding in organisational psychology and the scientific bedrock under every framework on the shelf: specific, difficult goals produce higher performance than vague or easy ones, across hundreds of studies and decades.
"Do your best" is measurably one of the worst instructions a leader can give.
The moderators are the machinery
The theory's moderators matter as much as its headline, because each one maps to a piece of OKR machinery. Commitment: difficulty only works when people accept the goal, which is why goals are written by teams and aligned, not cascaded. Feedback: goals without progress information do little, which is the scientific case for weekly check-ins and confidence assessment rather than set-and-forget annual objectives. Task complexity: the effect weakens as tasks become complex and novel, which is where the theory's most important refinement lives.
Learning goals versus performance goals
That refinement (Seijts and Latham): on complex, novel tasks, specific performance goals can actively hurt, because people grind on outcomes before they have discovered the method. Specific learning goals win there. This is the peer-reviewed backbone under two ZOKRI positions: discovery runs outside OKRs, as dual track discovery, and discovery Key Results were deleted from the methodology rather than graded.
The documented dark side
"Goals Gone Wild" (Ordóñez, Schweitzer, Galinsky, Bazerman, 2009) catalogued what happens when stretch goals meet incentives: tunnel vision, gaming, unethical behaviour, and the erosion of intrinsic motivation. That is the science under the bonus doctrine in OKRs and compensation: cash on team goals does not add motivation, it converts candour into arithmetic. The same evidence base underwrites aspirational targets and the grade-don't-score rule.
What we take, what we leave
We take specificity, feedback loops, team-owned commitment, and the learning-goal boundary. We leave the lab's preference for individual goals, because companies win or lose as systems: everyone rows in the same direction. The wider shelf this research underpins is mapped at Goal Setting.
Credited to Edwin Locke and Gary Latham; Seijts and Latham for the learning-goals refinement; Ordóñez, Schweitzer, Galinsky and Bazerman for the dark side. Machinery connections are ZOKRI methodology.
Does goal difficulty really improve performance? +
Yes, across hundreds of studies, provided commitment and feedback are present. That is why ZOKRI implementations pair ambitious targets with weekly check-ins and team-owned goal writing, the two moderators the research says make difficulty work.
When do stretch goals backfire? +
On complex, novel work (grind before discovery) and when wired to incentives (tunnel vision and gaming, per Goals Gone Wild). We design around both: discovery runs outside OKRs, and cash never touches team goal scores.
The science is settled; most implementations ignore it. We build the commitment, feedback and consequence design the research actually prescribes, with training and coaching that make it stick.