game-demo-feedback-triage（游戏演示反馈分流）

Game Demo Feedback Triage

Use this skill to turn a messy feedback packet into a short, weighted game-feedback brief.

The goal is not to summarize every note. The goal is to identify the highest-confidence player-facing blockers, separate representative signals from noisy anecdotes, and recommend the single next artifact that the team should produce before the next build, playtest, festival, or launch beat.

Read references/feedback-buckets.md before classifying edge cases or when store/demo, UX, stability, and performance signals are mixed together.

When to use this skill

• Internal playtest notes, survey responses, bug lists, and feedback docs need prioritization
• Steam Playtest, Next Fest, creator, or streamer demo reactions need to be converted into a ranked fix brief
• Teams have mixed qualitative and technical evidence and need to decide what matters before the next build
• Feedback touches onboarding, clarity, difficulty, controls, performance, stability, or conversion and the team needs one cross-functional view
• Small studios need a short go / no-go or fix-first recommendation instead of a giant product report

When not to use this skill

• The main problem is a raw build/log failure with no broader feedback packet → prefer game-build-log-triage
• The main problem is clearly performance capture and bottleneck diagnosis → prefer game-performance-profiler
• The team needs a store-page or launch-page audit before player feedback exists → prefer steam-store-launch-ops
• The user wants a generic sprint plan with no game-specific feedback evidence → prefer task-planning or bmad-gds

Instructions

Step 1: Label the feedback packet

Record the minimum facts before prioritizing.

Capture:

• source mix: internal playtest | Steam Playtest | creator/streamer | QA pass | survey/form | Discord/community | mixed
• build stage: prototype | vertical slice | demo | coming-soon | festival demo | launch candidate | post-launch
• audience fit: target players | friendly testers | broad audience | unknown
• evidence types: quotes, issue list, video/clip, survey counts, wishlist/context notes, perf/log findings
• target constraint: next playtest | Next Fest | creator beat | release date | unknown
• known severity signals: hard blockers, churn/bounce, confusion, low conversion, crashes, hitches, negative sentiment

If the packet is thin or heavily anecdotal, keep confidence low and say so.

Step 2: Cluster the feedback into one primary bucket set

Use these buckets and keep the list tight.

Primary buckets

• onboarding-comprehension
• core-loop-fun-clarity
• difficulty-balance-progression
• controls-ui-readability
• stability-performance
• content-polish-scope
• store-demo-conversion-message
• unknown-needs-better-evidence

Typical mappings

• "Players do not know what to do" → onboarding-comprehension
• "They get the controls but stop caring after 10 minutes" → core-loop-fun-clarity
• "The demo feels unfair / pacing is off / resource economy feels broken" → difficulty-balance-progression
• "Menu, HUD, tutorial prompts, or controller feel are frustrating" → controls-ui-readability
• "People mention crashes, hitches, Steam Deck issues, or stutter" → stability-performance
• "The build is mostly fine but rough edges or missing juice weaken the impression" → content-polish-scope
• "Players/viewers are not sold on the fantasy or demo/store pitch" → store-demo-conversion-message

Do not flatten every issue into "bugs" or "feedback." Separate player comprehension, fun, polish, stability, and conversion.

Step 3: Weight the signals before ranking

For each bucket, score the evidence using four lenses:

1. Frequency — how often did it appear across testers or sources?
2. Severity — does it block understanding, retention, or launch trust?
3. Representativeness — is it coming from target players or a skewed audience?
4. Timing risk — will ignoring it damage the next event, demo beat, or release?

Then mark each item as one of:

• fix-before-next-build
• watch-next-test
• do-not-overreact-yet

A loud single comment is not automatically a top priority.

Step 4: Build the feedback brief

Return a concise report with this exact structure:

# Game Demo Feedback Brief

## Scope
- Source mix: ...
- Build stage: ...
- Audience fit: ...
- Target constraint: ...
- Confidence: high | medium | low

## Highest-confidence signals
- 2-4 bullets on what the evidence most strongly suggests

## Priority buckets
| Bucket | Why it matters now | Evidence weight | Action |
|--------|---------------------|-----------------|--------|
| ... | ... | high / medium / low | fix-before-next-build / watch-next-test / do-not-overreact-yet |

## Top 3 fixes before the next beat
1. ...
2. ...
3. ...

## Issues to watch, not overreact to
- ...
- ...

## Evidence gaps
- ...
- ...

## Recommended next artifact
- Choose one: onboarding fix brief | demo polish checklist | creator-feedback summary | store/demo message update | performance follow-up brief | bug bash list | next playtest plan

## What not to do yet
- 1-3 bullets that prevent noisy backlog churn or premature redesigns

Step 5: Tailor the brief to the request type

For internal playtests:

• prioritize comprehension, fun, and progression before long-tail polish
• separate team-known issues from genuinely new player confusion

For Steam Playtest / Next Fest / public demo feedback:

• weigh reputational risk more heavily
• distinguish demo-quality blockers from store-page or traffic issues
• call out whether the next best move is a demo fix, a message fix, or a timing decision

For creator / streamer feedback:

• focus on first-session clarity, readability, showcase moments, and where the run becomes unwatchable or unshareable
• separate viewer-facing messaging issues from deep-system balance complaints

For mixed bug + perf + UX packets:

• use stability-performance as a bucket, not the whole conclusion
• if detailed profiling is needed, recommend game-performance-profiler as the next artifact, then fold the result back into the brief

Step 6: Ask for the minimum missing evidence when needed

If confidence is low, request the smallest next packet that would materially improve the triage:

1. 5-10 representative feedback quotes or issue bullets
2. whether the testers are target players or friendly/internal testers
3. the current build stage and next external milestone
4. any crash/perf pattern or device-specific complaint
5. one short summary of where players bounced, got confused, or lost interest

Do not ask for a giant postmortem.

Output format

Always return a short triage brief, not a long design essay.

Required qualities:

• prioritize the single strongest player-facing blocker first
• separate representative patterns from anecdotal outliers
• keep the report under roughly 350-550 words unless the user asks for more
• recommend one next artifact, not ten parallel workstreams
• keep launch/festival timing risk explicit when public-facing beats are involved

Examples

Example 1: Next Fest demo confusion

Input

We ran a small playtest on our Next Fest demo. Most people said the controls felt okay, but a lot of testers got lost in the first ten minutes and some streamers said they were not sure what the real hook was. A few people also mentioned hitches on Steam Deck.

Output sketch

• Primary bucket: onboarding-comprehension
• Secondary bucket: store-demo-conversion-message and stability-performance
• Top 3 fixes focus on early clarity, first-session signposting, and a targeted Steam Deck follow-up rather than a full redesign
• Recommended next artifact: onboarding fix brief

Example 2: Mixed internal notes before creator outreach

Input

Turn these playtest notes into priorities before we send the demo to creators. The bug list is long, but the real pattern seems to be that people love the combat and then bounce in the menus.

Output sketch

• Highest-confidence signal: core loop is landing but menu/control/readability is hurting continuation
• Priority action: fix high-friction menu/HUD readability issues before polishing low-frequency bugs
• Recommended next artifact: demo polish checklist or creator-feedback summary

Example 3: Hard feedback packet with store/demo confusion

Input

Our Steam Playtest signups are decent, but players either churn after the tutorial or say they expected a different kind of game from the page and trailer. What should we fix first?

Output sketch

• Buckets: onboarding-comprehension and store-demo-conversion-message
• Brief distinguishes expectation-setting from gameplay onboarding
• Recommended next artifact: store/demo message update or onboarding fix brief
• What not to do yet: do not bury the issue under broad scope expansion

Best practices

1. Treat feedback as evidence, not votes — prioritize representative blockers over loud anecdotes.
2. Separate comprehension from fun from stability — those produce different next artifacts.
3. Keep the next external beat visible — Next Fest, creator outreach, and launch windows change the weighting.
4. Use one short priority brief instead of flooding the team with every issue at once.
5. Route technical deep dives out when necessary — detailed profiling and raw log triage belong in their specialized skills.
6. Protect the backlog from churn — mark some issues as watch-next-test rather than immediate fixes.

References

• Steam Playtest
• Steam Next Fest - Tips
• Store Page, Building and Editing
• Best Practices: Five Tips for Better Playtesting
• 6 steps to a successful playtesting process for an indie developer
• Making Better Game Decisions with Feedback Testing
• How To Do An Efficient Game Test