Feedback4 min read

Why Feedback Backfires More Than a Third of the Time

Feedback lifts performance on average but makes it worse over a third of the time, mostly when it targets the person instead of the task. Here is the evidence.

UpTicker Team

The short answer

Feedback is not reliably good for you. It is reliably variable.

Kluger and DeNisi reviewed more than 600 effect sizes across roughly 23,000 observations and found that, on average, feedback interventions improved performance. But over a third of feedback interventions made performance worse. Not neutral. Worse than no feedback at all.

The pattern in the failures is consistent enough to be actionable. Feedback tends to help when it directs attention to the task. It tends to hurt when it directs attention to the self.

The finding most feedback advice ignores

The 1996 review in Psychological Bulletin is one of the most cited papers in the field, and one of the most selectively quoted. The average positive effect gets referenced constantly. The one third failure rate gets left out, which is unfortunate, because the failure rate is the useful part.

An average effect tells you what to expect if you pick a feedback intervention at random. Nobody wants to pick at random. What you want to know is which ones land in the harmful third, and the paper answers that.

Feedback Intervention Theory, in plain terms

Kluger and DeNisi's explanation runs on attention. You have a hierarchy of processes running while you work:

  • Task learning processes: how to do the thing better. Low level, detailed, method focused.
  • Task motivation processes: how much effort to put in. Mid level.
  • Meta task processes: what this says about me. High level, self focused.

Feedback moves attention somewhere on that hierarchy. Attention is finite, so whatever it moves toward, it moves away from something else.

Feedback that lands on task learning or task motivation generally improves performance. You now know what to change, or that you need to push harder.

Feedback that lands on meta task processes, meaning the self, generally does not. Attention leaves the task and goes to identity management: defending it, repairing it, or ruminating on it. Performance drops, and this holds for praise as well as criticism, because praise aimed at the person is still aimed at the person.

This is why "you are so talented at this" and "you are just not detail oriented" fail for the same structural reason, despite feeling completely different.

The practical rules that fall out of this

Feedback that tends to harmFeedback that tends to help
You are disorganisedThree of the five sections had no source attached
You are a natural at thisThe second paragraph carried the argument; the third repeated it
You did not try hard enoughThe draft went out without the pricing table, which is what the client asked for
Great job, you are crushing itResponse time dropped from 4 days to 1 day this week
You are falling behindYou are 18 sessions short of the pace needed to hit the target

Four rules:

  1. Describe the work, never the worker. The subject of the sentence should be the artefact or the behaviour, not the person.
  2. Attach a next action. Feedback without a corrective path pushes attention upward to the self by default, because there is nowhere else for it to go.
  3. Avoid comparison to other people. Normative comparison is one of the most reliable ways to move attention to identity.
  4. Be careful with praise. Person praise is not a safe alternative to person criticism. It has the same attentional problem.

Why UpTicker's coach never grades your character

This is a hard constraint in how UpTicker writes to you, not a stylistic preference.

The coach reports on the work and the numbers. It will tell you that three high priority items rolled over for the second consecutive day, that your completion rate on morning blocks is materially higher than afternoon blocks, or that you are behind the pace required for your named target. Every one of those statements has a task level next action attached to it.

It will not tell you that you are undisciplined, inconsistent, a procrastinator, or the sort of person who does not follow through. It will also not tell you that you are amazing, a machine, or built different. Both categories are meta task feedback, and the evidence says both categories move attention off the task.

The design test is simple: if a sentence would still make sense as a description of who you are after the app is deleted, it should not be in the app.

Frequently asked questions

Does feedback improve performance? On average yes, but Kluger and DeNisi (1996) found that over a third of feedback interventions reduced performance relative to no feedback. The average conceals a large harmful minority.

Why does feedback sometimes make things worse? Their Feedback Intervention Theory attributes it to attention. When feedback directs attention to the self rather than the task, attention is consumed by identity management instead of the work.

Is positive feedback always safe? No. Praise directed at the person rather than the work has the same attentional problem as person directed criticism. Praise the output and the method.

What does good feedback look like? Specific, task focused, descriptive of the work rather than the person, and paired with a concrete next action.

Should I stop giving people feedback? No. Change what it points at. Feedback about the task and how to improve it is the version supported by the evidence.

Reference

Kluger, A. N., and DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2), 254 to 284.

Turn the research into daily action.

UpTicker connects your goals, daily plan, next best action and progress in one execution system.

Download UpTicker