Goal setting

Goal-setting theory is one of the most replicated findings in applied psychology, and one of the most carelessly applied.

The finding, from Edwin Locke in 1968 and hundreds of studies with Gary Latham after: specific, difficult goals produce higher performance than vague ones, including the vaguest and most common of all, do your best. "Do your best" fails because it has no edge — any outcome can be reconstructed afterwards as your best, so there is nothing to aim at and nothing to fall short of.

That much is solid. What almost never survives the trip into a management deck is that Locke and Latham spent thirty years also documenting when it does not work — and the conditions are not footnotes. They decide the sign of the effect.

Locke and Latham's 2002 summary names the moderators. All four have to hold.

Commitment. The person has to actually accept the goal. An assigned target that is privately regarded as fantasy produces nothing, or worse than nothing. Commitment rises when the goal's importance is explained and when the person believes it is reachable — which is why stretch and absurd are separated by a line that the person being stretched gets to draw.

Feedback. A goal without progress information is a wish. This is the most commonly violated condition in real workplaces: an annual number with no visible monthly reading is functionally the same as no goal at all, because there is no way to adjust.

Task complexity. The effect is strongest on simple, well-understood tasks and weakest — sometimes reversed — on complex novel ones. On a task where you do not yet know how, a hard outcome goal adds pressure to a problem that needs exploration, and pressure narrows attention exactly when it should widen.

Ability and resources. A specific hard goal on a task you cannot yet do does not summon the capability. It produces a sequence of failures that reliably damages self-belief.

Read together, these say something unfashionable: goal setting is a technique for execution, not for discovery. For work whose method is already known, it is excellent. For work where the method is the problem, it is the wrong instrument.

The fix Latham himself proposed for the complexity problem is worth more than any acronym.

A performance goal names an outcome: raise conversion by 15%, ship by March, run it in under 45 minutes. A learning goal names an amount of discovery: find and test five distinct causes of drop-off, identify three approaches other teams use and try two.

Seijts and Latham's 2005 review of the evidence is clear about the split. On a task the person already knows how to do, performance goals win. On a novel or complex task, learning goals outperform them — sometimes substantially — because the useful early behaviour is search, and an outcome target suppresses search in favour of whatever is already known to move the number.

This is the most practically useful distinction in the literature, and the one to reach for when a goal is producing anxiety rather than action. The anxiety usually means an outcome goal has been set on a problem you do not yet have a method for. Converting it to a learning goal keeps the specificity, which is the part that works, and moves the difficulty to the part you control. The Growth Mindset test reads the belief that makes this conversion easy or hard.

In 2009 Ordóñez, Schweitzer, Galinsky and Bazerman published a deliberately pointed paper — Goals gone wild — arguing that goal setting is over-prescribed and that its side effects are systematic rather than incidental. Locke and Latham replied in the same issue; the exchange is worth knowing about, because the failure modes named are ones most people have watched happen.

Tunnel vision. A goal directs attention to the measured thing, which means attention leaves the unmeasured things. That is not a misuse of goal setting; it is the mechanism working. The safety, quality or relationship that was not in the goal is where the cost lands.

Horizon shortening. A quarterly target makes decisions that pay off in eighteen months harder to defend, even to yourself.

Ethical erosion. Several experiments find that specific hard targets, especially near-miss ones with rewards attached, increase misreporting. People a hair under the number round up, and in organisations this compounds.

Risk in the wrong direction. Behind on a goal, people take larger risks to catch up — a pattern that connects directly to what the entry on sunk cost in careers describes.

Goals as pressure. Where a goal has been internalised as a statement about your worth, missing it is not information, it is a verdict, and the machinery in the perfectionism entry takes over from there.

Four things follow, and none of them is an acronym.

Match the goal type to what you do not know. If you know the method, set the outcome. If you do not, set the search. Getting this backwards is the single most common self-inflicted goal injury.

Set the feedback before the target. If you cannot say how you will know where you stand in two weeks, you have a wish. Fixing the reading is usually more valuable than adjusting the number.

Keep one goal for one horizon. Multiple simultaneous hard goals do not multiply effort; they get traded against each other, and the one with the clearest measure wins regardless of which matters most.

Write down what you are agreeing to ignore. Tunnel vision is the cost of the technique, not a malfunction. Naming what is being deprioritised for the quarter — and what would count as having gone too far — is the cheapest available protection, and it is what separates a goal from a promise you will regret.

And where a personal goal is really a proxy for something else — status, proof, a verdict on whether you are enough — no amount of specificity helps. The entries on status anxiety and work and identity are the ones for that, and the Career Values test is a faster read on whether the goal is yours.