360-degree feedback collects ratings from four perspectives — manager, peers, direct reports, and self. It's exceptionally effective at developing management skills and exposing self-perception gaps, but bad operating design turns it into theater or breeds conflict.
This article isn't only a question template. The center of the article is the design principle: "observed behavior beats abstract ratings."
What 360 feedback is actually for
| Purpose | Importance |
|---|---|
| Surface self-perception gaps | ◎ Most important |
| Management skill development | ◎ |
| Build buy-in around evaluations | ○ |
| Drive comp / promotion decisions | △ Use cautiously |
The most important purpose is surfacing self-perception gaps. Tying it to comp turns raters cautious and kills the real value.
Six categories of questions
Category 1: Leadership
1. "Communicates org vision and direction clearly to the team"
2. "Leads the team forward, even in difficult situations"
3. "Makes key decisions at the right time"
4. "Supports team member growth with a long-term perspective"
Category 2: Communication
1. "Listens through to the end of what people say"
2. "Communicates their own thinking in unambiguous terms"
3. "Delivers negative feedback without injury"
4. "Takes team member feedback in good faith"
Category 3: Team building
1. "Facilitates collaboration between team members"
2. "Plays to each individual's strengths"
3. "Handles conflict and friction appropriately"
4. "Maintains psychological safety in the team"
Category 4: Execution
1. "Balances deadline and quality"
2. "Prioritizes appropriately"
3. "Delegates well without overloading themselves"
4. "Allocates resources appropriately"
Category 5: Member development
1. "Deliberately creates growth opportunities for team members"
2. "Provides specific, constructive feedback regularly"
3. "Supports career direction per individual"
4. "Treats failure as a learning opportunity"
Category 6: Self-management
1. "Controls emotions and stays composed"
2. "Maintains decision quality under stress"
3. "Recognizes their own weaknesses and works on them"
4. "Takes initiative on new challenges"
4 questions × 6 categories = 24 questions — the standard volume.
The hard part — from "rating" to "observing behavior"
Why ratings fail
✗ "Rate leadership on a 5-point scale"
Rater answer: "3 or 4"
→ No clarity on what to improve
Abstract concepts like "leadership" on a 5-point scale produce answers that don't point to anything actionable.
Behavior-observation design
○ "Communicates org vision and direction clearly to the team"
→ About a concrete observable behavior
→ Improvement direction is visible
○ "Delivers negative feedback without injury"
→ Observable behavior
→ Skill development direction is clear
Subjects shift from abstract concepts to observable behaviors. That's the substance.
Frequency-based scale
✗ 5-point: 1 (very poor) → 5 (very good)
○ 5-point:
1: Almost never observed
2: Sometimes observed
3: Somewhat observed
4: Frequently observed
5: Consistently observed
"Frequency" is easier for the rater to judge and clearer for the ratee to act on than "good / bad."
Open text design
Always include open text to capture specific facts that rating scales miss.
Recommended open-text questions
1. "Tell me about this person's strongest strength, with a specific story"
2. "What could this person improve? Share a specific situation"
3. "What kind of support would help this person grow?"
4. "Share a memorable episode from working with this person"
Asking for specific episodes yields concrete facts, not abstract impressions.
Rater selection
Standard composition
Self: 1
Manager: 1–2
Peers: 3–5
Reports: 3–5
Total: 8–13
Selection principles
- Only people with relevant working interaction
- Don't pack in people with strong personal stakes
- Diverse perspectives represented
- Tell the ratee why each rater was chosen
Optimal rater count
- ≤ 5: identification risk is high, candid feedback drops
- 6–10: standard, balanced
- 15+: management overhead high, rater fatigue risk
Anonymity discipline
The most critical element of 360 is candid feedback. Without anonymity, you only get diplomatic ratings.
Anonymity design
- Raters cannot choose named vs. anonymous — universally anonymous
- Ratees see aggregate results only, never individual responses
- Open text masked by an editor before being shown to the ratee
- Categories with fewer than 5 raters don't get aggregated (avoid identification)
When anonymity breaks
If you label "manager's comment" vs. "report's comment" and one category has only 1 person, that person is identifiable.
Counter: roll up manager, report, peer into a single "surrounding" label.
Delivering feedback
Handing over a numbers-only report is risky. Ratees take it hard or misread it.
Recommended delivery
1. Feedback meeting (1–1.5 hours)
2. Manager or HR partner present
3. Lead with overall pattern: "Strengths in area X, growth opportunity in Y"
4. Detail the strengths (including specific episodes)
5. Detail the improvement areas (constructive framing)
6. Have the ratee interpret and write their own development plan
7. Schedule follow-up
Not "hand over the results and done" — design through to the ratee acting on it.
Common 360 problems
Problem 1: Perspective divergence
Sometimes "managers rate high, reports rate low." This divergence can confuse the ratee.
Counter:
- Don't just hand over the divergence
- Reframe as "different perspectives revealing different aspects"
- Discuss in the feedback meeting
Problem 2: Popularity contests
Peer ratings tend to reward "well-liked" people. Likability and work quality are different.
Counter:
- Behavior-grounded questions reduce emotion
- Open text asking for concrete episodes
- Treat results as reference; final judgment with the manager
Problem 3: "Can't write honestly" pressure
Peers and reports avoid honest answers fearing identification.
Counter:
- Strong anonymity guarantees
- Explicit purpose: "feedback is for development"
- Zero historical incidents of "criticism traced back to its source"
Problem 4: Bad timing
Busy season, right before performance reviews, right after a reorg — fair rating becomes difficult.
Counter:
- Run during calmer periods
- Separate from performance-review (comp) timing
Frequency
| Cadence | Pro | Con |
|---|---|---|
| Annual | Lower workload | Slow improvement cycle |
| Biannual | Balanced | Rating fatigue caution |
| Quarterly | Faster improvement | High workload, theater risk |
Annual + lightweight pulse surveys is the realistic combination.
Tooling
360 is survey + management process combined.
Required features
- Per-rater unique URLs
- Anonymity (individual responses invisible to the ratee)
- Aggregate visualization (per category, per rater group)
- Open-text consolidation + masking
- Auto-generated reports per ratee
- Time-series comparison
Where Repoan fits
Repoan supports no-code 360 operation:
- Ratee-rater mapping — bulk configuration
- Auto-issue unique URLs — one per rater
- Auto anonymity — categories with n<5 excluded from aggregation
- Auto-generated radar charts — visualize per category
- AI open-text masking — auto-detect identifying details
- Per-ratee auto-generated report — feedback-meeting material
- Time-series comparison — track improvement over rounds
Summary
360 feedback design that works:
- Purpose: "surface self-perception gaps" and "skill development" — use comp decisions cautiously
- Design around observable behavior, not abstract ratings
- Use frequency, not "good/bad," on the scale
- Open text for specific episodes
- Anonymity is non-negotiable
- Design through the feedback meeting, not just the survey
- Integrate with 1-on-1s and daily management
"We do 360s" isn't the goal. "How the results change behavior" is the substance.