Goal-Setting Theory: What Locke & Latham Actually Found

"Set specific, challenging goals." It's management gospel—one of the few principles that seems universally accepted.

And it's true. Edwin Locke and Gary Latham's goal-setting theory, developed over decades, is one of the most validated findings in organizational psychology. Specific goals outperform vague goals. Challenging goals outperform easy goals.

But the research comes with caveats that most organizations ignore—and getting them wrong can actually backfire.

What the Theory Actually Says

Locke and Latham's core findings, validated across hundreds of studies:

Specificity matters. "Improve customer satisfaction" produces worse results than "Increase NPS from 45 to 55." Vague goals leave too much room for interpretation and don't create clear targets.

Challenge matters. Easy goals don't motivate. Challenging goals—stretch targets that require real effort—produce higher performance than "do your best" instructions.

Commitment matters. Goals only work when people actually accept them. Imposed goals without buy-in don't produce the same effects as goals people commit to.

Feedback matters. Goals without progress information don't drive improvement. People need to know how they're tracking.

So far, so intuitive. But here's where organizations often go wrong.

The Boundary Conditions

Goal-setting theory has important boundary conditions—situations where the standard advice doesn't apply or can backfire.

Task complexity. For complex tasks, learning goals outperform performance goals. If someone is figuring out how to do something, a specific performance target can actually harm outcomes by focusing on the end result rather than skill building.

Goal conflict. Multiple specific goals can undermine each other. If someone has ten specific targets, they'll optimize for the ones that feel most measurable—which may not be the most important.

Gaming. Specific goals invite gaming. "Make 50 calls per day" can be achieved by making 50 bad calls. The specificity that drives focus can drive narrow optimization at the expense of broader objectives.

Success criteria. Here's the one that matters most: specific goals without clear success criteria aren't actually specific. "Increase NPS by 10 points" is only meaningful if we've defined what achieving it looks like at different levels—and agreed on that definition before the period starts.

The Real Problem: Goals Without Criteria

I've reviewed thousands of goals in performance management systems. Most could be strengthened by adding clearer success criteria.

"Launch new product by Q3" sounds specific. But what does success look like?

• Launched with full feature set vs. MVP?
• On budget vs. over budget?
• Positive customer feedback vs. any feedback?
• First sale vs. revenue target?

Without these criteria, evaluation becomes subjective. The manager decides at year-end whether the goal was "met" based on their interpretation—which can reintroduce the subjectivity we're trying to reduce.

Goal Attainment Scaling, a methodology developed in 1968 and validated extensively in clinical and rehabilitation settings, addresses this directly. While formal GAS research has focused primarily on healthcare contexts, the underlying principle—defining observable success criteria at multiple levels before the period starts—aligns with decades of workplace goal-setting research.

In the GAS methodology, for every goal, you define what "exceeds," "meets," and "below" expectations look like—specifically and observably—before the period starts. Evaluation becomes verification: "Did they achieve the pre-defined level?" rather than "How do I feel about what they accomplished?"

The Cascade Problem

BCG's research uncovered another failure mode: leaders two to three levels below the C-suite frequently cannot articulate how their department contributes to company goals.

Goals cascade down beautifully. Strategic priorities become business unit OKRs become team objectives become individual goals. The hierarchy looks logical.

But the math rarely works back up. If every team hits their goals, does the company hit its objectives? Many organizations can't answer that question—and discover the gaps in December, not January.

This isn't a goal-setting problem per se. It's an alignment verification problem. Goals that look connected often aren't. And the cascade process, done without validation, can produce a portfolio of individual achievements that don't sum to organizational success.

What Goal-Setting Research Supports

Setting this all together, what does goal-setting research actually support?

A small number of clear priorities. Five to seven focused goals, not twenty-five metrics weighted at various percentages. The research says focus beats comprehensive measurement.

Success criteria defined upfront. Not "specific goals" in the sense of word count, but goals where everyone agrees what success looks like before the work begins.

Employee participation. Goals that people help set produce better commitment than goals imposed from above.

Regular progress check-ins. Goals without feedback don't drive improvement. Frequent conversations about progress matter.

Alignment verification. Goals that cascade down should add back up. Test the math before the period starts.

Flexibility when conditions change. Goal-setting theory says commit to goals. But rigid commitment to goals that no longer make sense—because the market shifted or priorities changed—produces worse outcomes than sensible adjustment.

Try This

Pull up your organization's goal-setting guidance. Ask:

• Do we require specific success criteria at multiple levels (not just the target)?
• Do employees participate in setting their goals, or receive them?
• Do we verify that cascaded goals add back up?
• Can goals be adjusted when business conditions change?
• Do we focus on a small number of priorities, or comprehensive measurement?

If the answers reveal gaps, you're not alone. Most goal-setting processes honor the letter of "specific, challenging goals" without the conditions that make them work.