The annual review is too late to be useful as feedback and too important to be treated as a formality. Organizations should keep it for pay, promotion, calibration, and a formal record, but move developmental feedback into small rituals tied to real work. That split—not a larger review or a mandate to comment on everything—is the foundation of a feedback culture that produces continuous feedback and better performance conversations.
The strongest argument for the annual review is not nostalgia. It gives managers a shared checkpoint, creates a documented account of performance, and can reduce the unfairness of one manager making decisions from scattered impressions. Frequent feedback can feel like surveillance. Different managers may apply different standards. Poorly delivered comments can reinforce bias, while a compensation cycle still needs one official narrative. Those are serious concerns, especially in large organizations.
They are also reasons to keep the annual review, not reasons to make it carry the whole burden of development. A yearly meeting is useful for looking backward and making formal decisions. It is a poor mechanism for helping someone change what they do on their next client call, project handoff, or team meeting.
There is a less comfortable reason to resist the idea that more feedback is automatically better. Kluger and DeNisi’s 1996 meta-analysis of 131 feedback interventions found that feedback improved performance on average, but more than one-third of the interventions reduced performance. The harmful cases were more likely when feedback pulled attention toward the self—How am I doing?—instead of the task—What should I do next?
The point is not to avoid feedback. It is to design feedback so that it produces a specific next action rather than a lingering judgment about someone’s character or worth.
Keep the annual review in its lane
A useful performance system separates three activities that are often blended together. Feedback describes what happened and what effect it had. Coaching helps someone decide what to try next. Evaluation makes a judgment about performance against a role, level, or standard.
One conversation can include all three, but people should know when the conversation has shifted. A manager who begins with a developmental observation and then quietly turns it into evidence for a promotion decision has changed the stakes without saying so. The employee will quite reasonably stop treating the exchange as a safe place to test an adjustment.
The annual review has a legitimate evaluation job. It can gather patterns from the year, clarify priorities, discuss pay or promotion, and record decisions that should not be left to memory. Developmental feedback belongs closer to the work. When it is delayed until the formal review, the employee has often repeated the behavior dozens of times, and the manager has accumulated frustration instead of usable evidence.
That is why continuous feedback should not mean constant commentary. It means a reliable connection between an observable event, a useful response, and another chance to apply what was discussed. The annual review becomes a summary of that work rather than the first time anyone mentions a problem.
Build the ritual around a real piece of work
Start with an event, not a personality label. Center for Creative Leadership’s Situation-Behavior-Impact model is a practical guardrail: name the situation, describe the behavior, and explain the impact. It keeps a manager from saying that someone is careless, passive, or difficult when the available evidence is a particular action in a particular setting.
The model needs one more step to become a learning ritual: an agreed experiment and a date to revisit it. A manager might say:
On Tuesday’s renewal call, when the client paused, you answered before they finished. The pricing concern only surfaced at the end, so we lost the chance to address it early. On the next call, let a full pause pass before responding. How does that observation fit with your read of the conversation, and what would help you try it?
The employee needs room to add context, disagree with the interpretation, or identify a different cause. The manager’s observation may be accurate while the proposed solution is wrong. The exchange is useful when both people leave with a behavior that can be observed, not when the manager has delivered a polished verdict.
Keep a short record. A one-page feedback loop can include the date and work event, the behavior observed, its impact or relevant standard, the employee’s response, the next behavior to test, and the revisit date. That note is working memory, not a secret dossier or a numerical score. If the issue later becomes a formal performance matter, use the organization’s proper process rather than quietly converting informal notes into an undisclosed case file.
One behavior per conversation is usually enough. A manager who delivers six corrections at once may feel thorough, but the employee leaves with no clear experiment. Specificity is more demanding than volume.
Set the cadence around the work, not the calendar
How often should continuous feedback happen? There is no universal weekly quota. Use the shortest interval that gives a person another opportunity to apply the advice without turning the manager into a running commentary on every move.
A workable team ritual has three natural triggers:
- after a consequential piece of work, such as a customer escalation, project milestone, presentation, or safety-critical task;
- when a behavior becomes a pattern rather than a one-off mistake; and
- before the next comparable opportunity, so the person can use the advice while the context is still familiar.
These triggers work better than a blanket instruction to provide feedback every Friday. A customer-support supervisor may debrief after a difficult escalation. A project lead may hold a short review after a launch milestone. A manager of managers may need a monthly pattern conversation because the work is less frequent and less observable. The cadence should follow the work’s learning cycle.
The cost is real. For a manager with eight direct reports, a 12-minute weekly loop takes 96 minutes before preparation and follow-up. That may be worthwhile for a high-risk or fast-moving team and excessive for independent work with few meaningful handoffs. Biweekly or event-based conversations can be more honest than a weekly ritual that produces invented observations.
Do not turn every one-to-one into a critique session either. People need status checks, recognition, problem-solving, and ordinary human conversation. If every meeting carries the threat of correction, employees will protect themselves by saying less. Recognition can belong in the ritual, but it should name the behavior that made the result possible rather than offering a vague “great job.”
Make the loop two-way and consequential
A feedback ritual without a changed behavior is theater. The manager’s job is not finished when the observation has been delivered; the employee’s job is not finished when they have nodded politely.
Leaders should go first by asking for feedback on something they can change soon. “What did I do in the last planning meeting that made the decision harder?” is more useful than “Do you have any feedback for me?” The first question gives the other person a time, behavior, and possible impact to consider. The leader then has to do something visible with the answer.
Amy Edmondson’s research on psychological safety gives the relevant test: can people take an interpersonal risk, such as admitting a mistake or challenging a manager, without expecting punishment? A branded script cannot create that condition. A manager who asks for candor and then argues with every answer will teach the team that the ritual is decorative.
The same principle applies to peer feedback. Make the request narrow and voluntary where possible. A colleague can ask, “What is one thing I should repeat and one thing I should change in the handoff?” That is easier to answer than an invitation to assess a person’s entire working style.
Measure the loop by what happens next, not by the number of feedback entries. Did the conversation occur close enough to the event to be useful? Did the person choose an observable next behavior? Was there another opportunity to try it? Did the same issue recur, and if so, was the cause a skill gap, an unclear standard, or a broken process?
Those questions protect against a common failure in feedback culture: rewarding managers for producing documentation instead of improving work. A hundred logged conversations can represent a very busy form of avoidance if no one checks whether behavior or conditions changed.
Know when feedback is the wrong tool
Feedback rituals fail when they are used to solve problems caused by the system. If a service representative is told to slow down with customers while a compensation plan rewards shorter calls, the issue is not simply conversational skill. If a project manager is criticized for missing dates while priorities change weekly and dependencies remain unresolved, another private performance conversation may be scapegoating.
Pause and clarify the standard before correcting the person. Check workload, authority, tools, incentives, and conflicting instructions. Sometimes the right response is a process change or a role decision, not a better sentence from a manager.
This advice also does not apply as a substitute for formal action in cases involving harassment, discrimination, safety, fraud, serious misconduct, or a sustained performance problem that requires a documented plan. Follow organizational policy and involve the appropriate HR or employee-relations partner. A casual ritual should never be used to keep a serious issue informal.
Trust is another limit. In a team where people have seen feedback used for retaliation or surprise dismissals, announcing a new feedback culture will not fix the problem. Start with manager behavior: clear standards, accurate observations, prompt follow-through, and an explicit separation between developmental coaching and formal evaluation. Trust is earned through repeated consequences, not declared in a launch email.
A one-week test
Choose one recurring work moment before Friday: a client call, project handoff, shift debrief, or team decision. Tell the participants that the purpose is to improve the next attempt, not to create a score. Spend 12 minutes naming one observed behavior, its effect, the person’s interpretation, and one behavior to test. Write down the revisit date, then return to it after the next comparable event.
If the experiment produces a clearer action and a better follow-up conversation, repeat it. If it produces polite silence, defensiveness, or a made-up observation, fix the conditions before increasing the frequency. Done well, the next annual review will have less archaeology to perform—and more real evidence about how people learned to work differently.