Praise vs feedback: when each lands, when it backfires

Most managers treat praise and feedback as two settings on the same dial — praise is the nice end, feedback is the difficult end, and a good manager dispenses roughly the right ratio of each. That framing is why both get used badly. They aren't two temperatures of the same thing; they're two different instruments doing two different jobs. Praise tells someone this specific thing worked, do it again — it's reinforcement. Feedback tells someone here's the gap between what happened and what's needed — it's calibration. Swap them and you get the two failure modes every team recognizes: praise so vague it means nothing, and criticism dressed up as encouragement so it can be safely ignored.

Vague praise is worse than no praise

"Great work this week!" feels like a gift and functions as noise. The problem is precision: the person cannot repeat what they can't identify. If they don't know which behavior earned the praise, they'll keep doing everything they did — including the parts that didn't help — or, more commonly, they'll file it as pleasant background static and change nothing.

There's a second, less obvious cost. A manager who praises constantly and unspecifically devalues their own currency. Once praise is understood to be automatic, it stops carrying information, so genuine recognition of something genuinely good no longer registers. And there's a third: in a team where praise is reflexive and criticism is rare, praise becomes a signal of impending bad news — the manager who opens with "you're doing great" is heard as clearing their throat. That's the feedback sandwich problem in its purest form: the compliment has been trained into a warning.

Specific praise avoids all three, and it's the same discipline as specific feedback — a moment, a behavior, an effect:

When praise is the wrong tool

Praise reinforces. So it's the wrong instrument any time the actual need is calibration — and reaching for it anyway is usually conflict-avoidance wearing a friendly face.

• When there's a real gap you haven't named. Praising around a problem — complimenting the parts that worked while leaving the thing that didn't unmentioned — reads as approval. The person walks away believing the work was fine, which is a worse outcome than saying nothing at all.
• When it's used to soften a correction into invisibility. If the praise exists to make the criticism survivable, the criticism usually doesn't survive: people remember the compliment and discard the qualifier.
• When someone needs to know where they stand. Praise can't answer "am I meeting the bar?" Only calibration does. A stream of encouragement to someone who is quietly underperforming is one of the least kind things a manager can do, because it removes their chance to fix it while fixing it is still easy.
• When it's a substitute for something material. Recognition doesn't stand in for scope, pay, or a promotion, and using it that way is quickly and correctly read as cheap.

When feedback is the wrong tool

The reverse error is less discussed and just as real. Sometimes what's needed is reinforcement, and offering calibration instead is actively counterproductive.

• When someone did it right and needs to know it was right. After a hard piece of work that went well, "here's what to improve next time" is the wrong response — not because improvement doesn't exist, but because the useful information in that moment is which parts worked. Skip it and the person is left guessing whether they got away with something.
• When they're mid-attempt at a change you asked for. Someone actively working on a behavior needs to know when they're doing it correctly, or they can't tell progress from noise. Calibration on top of calibration, with no reinforcement, reads as a target that keeps moving.
• When the person is already over-critical of themselves. For someone whose self-assessment runs harsher than reality, more correction lands as confirmation of a story they're already telling — and the accurate, useful information is the specific evidence that they're doing better than they think. Which one you're dealing with is impossible to know from the inside, and it's precisely the humility-versus-impostor question.

Getting the ratio right without gaming it

The advice to praise more than you criticize is common, and directionally fine — people need reinforcement to stay oriented, and a diet of pure correction burns them out. Research from Gallup on recognition and engagement consistently finds that people who receive specific, regular recognition are more engaged and stay longer, and Knowledge at Wharton's work on motivation points the same way: reinforcement is not a soft extra, it's how people learn what to keep doing.

But treating that as a quota is where it goes wrong. A manager counting to a ratio starts manufacturing praise to earn the right to criticize, and manufactured praise is exactly the vague kind that devalues the currency. The workable version isn't a ratio at all: make both specific, and let frequency follow reality. In most healthy teams, most of what happens is good, so specific praise naturally outnumbers correction without anyone engineering it. If in your team the honest count comes out mostly negative, that's information about the situation — a hiring problem, an unclear standard, an overloaded team — and inflating the praise column won't fix it.

How to make both count

1. 1

   Name the behavior, not the person

   "That was smart" is a verdict; "walking through the tradeoff before recommending one made the decision easy to agree with" is reusable. This applies identically to praise and to correction — behavior over character in both directions.
2. 2

   Add the effect

   Say what the behavior caused: it kept the room calm, it saved a rework cycle, it made the decision defensible. The effect is what turns a compliment into information the person can act on.
3. 3

   Deliver them separately

   Don't bundle. If you have both to give, give them at different moments or at least clearly separated, so neither is heard as the setup for the other. Mixed in one breath, the praise gets discounted and the correction gets softened out of existence.
4. 4

   Ask which job you're doing before you speak

   Are you trying to get more of something (reinforcement) or to close a gap (calibration)? Naming it privately, in one second, prevents almost every misuse in this article — including the very common one of praising because you're avoiding the harder conversation.
5. 5

   Don't let praise live only in public and correction only in private

   Public praise is good, but if it's the only place praise appears it reads as performance. Specific praise in a 1:1 — where nobody's watching and there's nothing to gain — is often the version people actually believe.

Where Mirorly fits

Here's the asymmetry worth sitting with: as a manager you almost certainly receive far more praise than feedback, and almost all of it is the vague kind. "You're doing great, keep it up" — from your boss, who's too removed to be specific, and from your team, for whom honest calibration of their manager carries real risk. So the reinforcement you get is uninformative and the calibration mostly doesn't arrive at all, which leaves you unable to tell which of your behaviors to repeat and which to change. Mirorly closes both halves: your team answers specific behavioral questions about how you actually work — surfacing not just gaps but the things they'd want you to keep doing — aggregated with an anonymity threshold so candour is safe, and set next to your own answers so you can see where your read is off. The core leadership behaviours template is built for that, and knowing precisely what's working is every bit as actionable as knowing what isn't.

Common questions

The one-line summary

Praise and feedback are different instruments — reinforcement versus calibration — so using one to do the other's job produces vague praise that devalues itself and criticism soft enough to ignore; make both specific (a moment, a behavior, its effect), deliver them separately, ask yourself which job you're actually doing before you speak, and let frequency follow what's really happening rather than a ratio you're managing to.