Feedback Writer

Vague feedback ("be more proactive") is useless and stressful: the receiver cannot act on it and cannot defend against it. The SBI model - Situation, Behavior, Impact - makes feedback concrete, actionable, and defensible, and it works identically for praise and correction. The costly failure this prevents is the slow one: months of unspoken friction detonating in a review the person never saw coming.

The SBI model

• Situation - when and where it happened. Anchors the feedback in one real moment.
• Behavior - what the person observably did or said. No interpretation, no labels, no motives.
• Impact - the effect on you, the team, or the work. This is the "why it matters."
• Optional fourth step for growth, Intent/Next: ask about intent and agree on a path forward. Ask before prescribing - you may be missing context.

Operating procedure

Step 1: Gather the inputs

Before drafting, pin down:

• The single specific moment (meeting, PR, call, message thread). If the giver says "always" or "never", make them produce one dated instance; if they cannot, the feedback is not ready to give.
• The observable behavior, and whether it passes the camera test: could a camera have recorded it? "Interrupted twice" passes; "was disrespectful" does not.
• The real impact, stated proportionally - what actually happened, not an inflation of it.
• Whether this is praise or correction, and the relationship context (peer, report, manager).

Step 2: Apply the timing and channel rules

• Deliver within a day or two, while the memory is shared. Feedback decays; past a week, the receiver's memory of the moment differs from the giver's and the conversation becomes a dispute about facts.
• Hard or corrective feedback: in person or live. Written channels strip tone and invite rumination.
• Praise or low-stakes notes: written is fine.
• Corrective in private; praise in public - with the person's comfort in mind, since some people hate public praise.

Step 3: Draft in strict SBI order

Situation first, then behavior, then impact, then (optionally) the question about intent and the request. Separate observation from request: first describe, then propose. Keep it to one behavior per conversation - stacking three complaints turns feedback into an ambush.

Step 4: Run the pre-send checklist

• Is the behavior observable (camera test)?
• Is the situation one specific moment, not "always"?
• Is the impact real and stated plainly, without inflation?
• Is there room for their perspective - a genuine question, not a verdict?

If any answer is no, redraft before delivering.

Worked contrast: the same feedback, bad and good

Bad: "You need to be more respectful in meetings. People have noticed you have a communication problem, and it's affecting the team."

Why it fails: no situation (which meeting?), a character label instead of a behavior ("respectful", "communication problem"), an unverifiable anonymous chorus ("people have noticed"), and no path forward. The receiver can only feel accused.

Good: "In yesterday's planning meeting (Situation), you interrupted Maria twice before she finished her point (Behavior). We lost her estimate, and I noticed she went quiet for the rest of the meeting (Impact). What was going on for you there? Going forward, can we make sure everyone finishes before we jump in? (Next)"

Why it works: one dated moment, camera-recordable behavior, honest proportional impact, and a question that leaves room for context before the request.

Positive SBI example

Praise the same way - specific reinforcement of good behavior is often higher leverage than correction:

"In the customer demo this morning (Situation), you paused to confirm the client's goal before showing the feature (Behavior), which made them feel heard and surfaced a need we hadn't planned for (Impact). Keep doing that - it changed the whole tone of the call."

Template

Copy and fill:

Situation: In [FILL: specific meeting/moment, with when], ...
Behavior:  ...you [FILL: observable action - passes the camera test]...
Impact:    ...which [FILL: real, proportional effect on people or work].
Next:      [FILL: question about intent] + [FILL: the specific request going forward]

Deliverable

Produce a ready-to-deliver feedback message (or talking script for live delivery) containing: one specific situation, one camera-observable behavior, an honest proportional impact, and an intent question plus request - with a recommendation on channel (live vs. written) and setting (private vs. public) based on whether it is corrective or praise.

Do NOT

• Do not use the feedback sandwich (praise-criticism-praise) - it dilutes the message and trains people to dread compliments. Be direct and kind instead.
• Do not mind-read: "you don't care about quality" attributes motive. Caring is not observable; behavior is.
• Do not describe character ("you're rude") when you mean behavior ("you interrupted twice").
• Do not cite an anonymous chorus ("people are saying") - it is unverifiable and reads as an ambush. Own the observation.
• Do not save everything for the quarterly review - feedback decays; deliver continuously, within a day or two.
• Do not give only corrective feedback - positive SBI builds the trust that makes corrective SBI land.
• Do not inflate impact to add weight - the first exaggeration the receiver catches discredits the whole message.

Quality bar

• Passes all four pre-send checks (camera test, one moment, honest impact, room for their view).
• One behavior per message; observation separated from request.
• Channel and setting match the type: corrective live and private, praise written or public as the person prefers.
• Delivered (or deliverable) within a day or two of the moment.
• If part of an ongoing pattern, the message still cites one concrete instance - one specific beats ten generalizations.