The science of feedback
High-information feedback works four times better than praise
Wisniewski, B., Zierer, K., & Hattie, J. (2020). The Power of Feedback Revisited. Frontiers in Psychology, 10:3087.
Wisniewski, Zierer and Hattie revisited the feedback question with 435 studies and a sharper knife: they sorted feedback by how much usable information it carried. The result is stark. Feedback dense with information lands at d=0.99 - a large effect. Praise and reinforcement land at d=0.24. Telling someone something true and specific is roughly four times more powerful than telling them they did great.
Data table
| Item | Cohen's d |
|---|---|
| High-information feedback (what happened + what to do) | 0.99 |
| Low-information / praise (reinforcement, encouragement) | 0.24 |
What they did#
The team coded 435 studies not by how feedback felt but by how much it told you - did it describe the gap and a way to close it, or did it just react? Then they compared effect sizes across those bins.
What they found#
Information won, decisively. A d of 0.99 is the kind of effect that changes outcomes; a d of 0.24 barely moves the needle. Praise is not neutral filler, either - hollow praise dilutes the useful feedback sitting next to it, dragging the whole exchange toward the low number.
Warmth is the wrapper, not the gift. The gift is a true, specific, actionable observation.
What it means for Speech Away#
This is why our feedback is built from numbers and mechanisms, not adjectives. Not "great job" but "your point arrived at 0:38 - try leading with it so the listener has a frame for everything after." The encouragement is real, but it sits around the information, never in place of it.