Learn how to give constructive feedback that lands, using the Situation-Behavior-Impact model instead of the feedback sandwich, with real examples for managers.
TL;DR: The feedback sandwich fails once a report learns to recognize the pattern and starts discounting the opening praise; the Situation-Behavior-Impact (SBI) model from the Center for Creative Leadership gives managers a more honest, repeatable alternative for delivering feedback that actually gets heard.
How to give constructive feedback that actually lands starts with retiring the feedback sandwich. Praise, criticism, praise feels considerate to deliver, but once a report recognizes the pattern, they discount the opening compliment and simply wait for the correction buried in the middle. The fix isn't softer language, it's a different structure: Situation-Behavior-Impact (SBI), the framework built by the Center for Creative Leadership that separates the observable behavior from its effect instead of wrapping it in unrelated praise. This piece myth-busts the sandwich specifically, walks through SBI in practice, and covers how often to use it.
The praise-criticism-praise structure traces back further than most managers realize. According to Wikipedia's history of the technique, it was popularized in the 1980s by Mary Kay Ash, founder of Mary Kay Cosmetics, who advised managers to wrap critical remarks between layers of praise. The logic made sense on paper: soften the blow, keep morale intact, deliver the message anyway.
The problem shows up over time, not on first use. Research from Ivey Business School at Western University, led by Assistant Professor Karen MacMillan, found that once a manager uses the sandwich repeatedly, reports learn to recognize it as a pattern.
MacMillan explains it directly: "They're waiting for the other shoe to drop. Regularly starting with a compliment to ease into a critique quickly teaches people to be on high alert. They end up ignoring the positive and resenting the negative."
The same mechanism shows up in an independent interview with Newsweek, where MacMillan explains that "receivers start to ignore the positive as they wait for the other shoe to drop." The opening praise stops functioning as recognition and starts functioning as a warning sign. That is the opposite of what the sandwich was designed to do.
Harvard Business Review has documented the same failure from a different angle. Dane Jensen and Peggy Baumgartner write that softening tough feedback "makes correcting people easier for you, but it rarely achieves the goal of helping someone improve." Mixed messaging also lands differently depending on who receives it.
"For people with strong egos, the criticism goes down too easily, while highly self-critical employees can choke on it," Jensen and Baumgartner write. "Others can find the mixed messages merely confusing."
A separate HBR analysis of a Harvard Business School field study found that critical peer feedback did not reliably drive improvement either. Employees who got critical reviews from colleagues tended to adjust their networks to spend more time around people who would rate them positively, rather than fix the underlying behavior. The pattern in both cases is the same: people manage the feedback giver's opinion of them instead of engaging with the substance.
One controlled study complicates a blanket "the sandwich never works" claim, and it's worth naming directly. Prochazka, Ovcari, and Durinik (2020) tested sandwich feedback against corrective-only and no-feedback conditions with 91 university students on a math task, and the sandwich group solved more problems and prepared longer than the other two groups. That was a single, one-time exposure in a computer-mediated lab setting, not a manager and report working together over months. The failure mode MacMillan's team describes is specifically about repeated use, where the report has enough exposure to learn the pattern and stop trusting it.
SBI does not try to balance criticism with praise. It separates what happened from how it landed, using three distinct components defined by the Center for Creative Leadership:
Notice what's missing: no opening compliment unrelated to the issue, no cushioning adjective, no attempt to manage the report's mood before you've said anything real. The structure itself does the work that praise was supposed to do, because it's specific enough that the person can actually picture what you're describing.
CCL also offers an extended version called SBII, which adds a fourth step: asking about intent after you've named the impact. That closes the gap between what someone meant to do and what actually landed, which matters when the impact was clearly not what the person intended.
Steps five and six are where a lot of managers rush past, because rushing feels efficient. A monologue is not feedback, it's a lecture. A report who never gets space to respond has no reason to trust the conversation was two-directional.
This structure also answers a common worry: how to deliver negative feedback without it feeling like a personal attack. Because SBI sticks to observable behavior and its impact rather than character judgments, the report has something concrete to respond to instead of a vague sense of being criticized as a person. Whether they actually engage with it also depends on whether they feel safe enough to respond honestly, which is the whole premise behind psychological safety at work.
None of these examples include a compliment bolted onto the front to soften the delivery. The specificity is what makes them land, not the tone.
Criticism points at the person. Constructive feedback points at the behavior and its consequence, then leaves room for the person to fix it.
"You're not detail-oriented" is a judgment about identity, and there's nowhere for the person to go with it. "The report went out with three pricing errors, and the client called to ask which number was correct" is a fact and a consequence, and it gives the person something specific to change.
That distinction is also what separates real accountability from blame. Extreme ownership as a leadership principle applies here too: a manager who owns the outcome of a project is more credible delivering hard feedback about it than one who's only ever pointing at other people's mistakes.
Kim Scott's Radical Candor framework maps feedback across two axes: how much you Care Personally and how directly you Challenge. The quadrant where both are high is Radical Candor itself. The other three are failure modes: Obnoxious Aggression (challenge without care), Ruinous Empathy (care without challenge), and Manipulative Insincerity (neither).
The feedback sandwich is a textbook example of Ruinous Empathy dressed up as directness. Scott's own team makes this point on the company's blog, describing how people "discard the meat" (the actual criticism) "and instead focus on the two delicious pieces of... bread" (the praise). The structure protects the manager's comfort more than it serves the report.
This is also where radical truth and transparency as a broader team culture connects back to individual feedback conversations. A manager can't run one honest feedback conversation inside a culture where directness is otherwise punished or avoided. The two have to move together.
More often than the annual review cycle most companies still default to. Gallup found that just one in four employees strongly agree they receive valuable feedback, and the ones who do are five times more likely to be engaged, 57% less likely to burn out, and 48% less likely to be job-searching.
SHRM highlights research showing that four in five employees (80%) who say they received meaningful feedback in the last week are fully engaged. Waiting for a formal review cycle to say something that could have been said the week it happened wastes most of feedback's value, whether you're measuring it through engagement, burnout, or retention risk.
Knowing the SBI structure and actually delivering it under pressure are different skills. Most managers who avoid feedback conversations aren't confused about the framework, they're anxious about how the other person will react.
This is the specific gap an AI mentor is useful for. GPTnius lets a manager talk through a real, upcoming feedback conversation, situation, behavior, and impact included, before walking into the room, so the wording gets tested on a mentor instead of on the report. If you want to try that before your next one-on-one, start free with an AI mentor and rehearse the conversation you've been putting off.
One distinction worth making: this article is about a manager delivering feedback to someone else. If you're interested in the individual side, how AI-personalized feedback loops can build your own growth mindset over time, that's a related but separate topic covered in growth mindset training with adaptive feedback.
The feedback sandwich opens and closes with praise around a piece of criticism in the middle. Research from Ivey Business School found that once a manager uses it repeatedly, reports learn to recognize the pattern and discount the opening praise while bracing for criticism, which undermines the recognition it was meant to convey.
SBI is a feedback framework from the Center for Creative Leadership that separates what happened from how it landed. Situation names when and where an event occurred, Behavior describes the specific, observable action without judgment, and Impact states the effect it had, positive or negative. It replaces unrelated praise with something a person can actually act on.
Deliver it soon after the event while it's still fresh, name the situation in one sentence, then describe only the observable behavior and its impact on the work, team, or client. Pause so the employee can respond instead of turning it into a monologue, then agree together on what changes next time.
Kim Scott's Radical Candor framework is another practitioner alternative, built on caring personally while challenging directly instead of cushioning criticism with unrelated praise. Scott's own team has argued the sandwich fails because people discard the criticism in the middle and only remember the praise on either side of it.
More often than a once-a-year review. Gallup found only one in four employees strongly agree they get valuable feedback, and those who do are far more engaged and less likely to burn out or job-search. SHRM highlights research showing four in five employees who received meaningful feedback in the last week are fully engaged.