game-demo-feedback-triage

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Game Demo Feedback Triage

游戏Demo反馈分类整理

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.
使用该技能将杂乱的反馈包转化为简短、加权的游戏反馈简报。
目标并非总结每一条记录,而是识别可信度最高的玩家侧阻塞问题,区分代表性信号与嘈杂的轶事,并推荐团队在下一版本、Playtest、展会或发布节点前应生成的单个后续工件
在分类边缘案例,或商店/Demo、UX、稳定性和性能信号混合时,请先阅读references/feedback-buckets.md

When to use this skill

何时使用该技能

  • 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
  • 内部Playtest记录、调查反馈、Bug列表和反馈文档需要优先级排序
  • Steam Playtest、Next Fest、创作者或主播的Demo反响需要转化为排名修复简报
  • 团队拥有混合的定性和技术证据,需要决定在下一版本前哪些事项至关重要
  • 反馈涉及新手引导、清晰度、难度、操控、性能、稳定性或转化,团队需要跨职能视角
  • 小型工作室需要简短的“执行/暂停”或“优先修复”建议,而非冗长的产品报告

When not to use this skill

何时不使用该技能

  • 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
  • 主要问题是原始版本/日志故障,无更广泛的反馈包 → 优先使用
    game-build-log-triage
  • 主要问题明确是性能捕获和瓶颈诊断 → 优先使用
    game-performance-profiler
  • 团队需要在玩家反馈出现前对商店页面或发布页面进行审核 → 优先使用
    steam-store-launch-ops
  • 用户需要无游戏特定反馈证据的通用冲刺计划 → 优先使用
    task-planning
    bmad-gds

Instructions

操作步骤

Step 1: Label the feedback packet

步骤1:标记反馈包

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.
在优先级排序前记录基本信息。
需捕获:
  • 来源组合:内部Playtest | Steam Playtest | 创作者/主播 | QA测试 | 调查/表单 | Discord/社区 | 混合来源
  • 版本阶段:原型 | 垂直切片 | Demo | 即将推出 | 展会Demo | 发布候选版 | 发布后
  • 受众匹配度:目标玩家 | 友好测试者 | 广泛受众 | 未知
  • 证据类型:引用、问题列表、视频/片段、调查统计、愿望单/背景记录、性能/日志数据
  • 目标约束:下一次Playtest | Next Fest | 创作者节点 | 发布日期 | 未知
  • 已知严重程度信号:硬阻塞、用户流失/放弃、困惑、低转化、崩溃、卡顿、负面情绪
若反馈包内容单薄或多为轶事,需降低可信度并明确说明。

Step 2: Cluster the feedback into one primary bucket set

步骤2:将反馈聚类至一个主桶集合

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.
使用以下分类桶,保持列表简洁。
主分类桶
  • 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
典型映射
  • “玩家不知道该做什么” →
    onboarding-comprehension
  • “他们掌握了操控,但10分钟后就失去兴趣” →
    core-loop-fun-clarity
  • “Demo感觉不公平/节奏失衡/资源系统崩坏” →
    difficulty-balance-progression
  • “菜单、HUD、教程提示或控制器手感令人沮丧” →
    controls-ui-readability
  • “用户提到崩溃、卡顿、Steam Deck问题或掉帧” →
    stability-performance
  • “版本整体无大碍,但粗糙的细节或缺失的亮点削弱了印象” →
    content-polish-scope
  • “玩家/观众不认可游戏设定或Demo/商店宣传点” →
    store-demo-conversion-message
不要将所有问题归为“Bug”或“反馈”,需区分玩家理解度、趣味性、打磨度、稳定性和转化问题。

Step 3: Weight the signals before ranking

步骤3:在排序前加权信号

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.
针对每个分类桶,从四个维度为证据评分:
  1. 频率 —— 在测试者或来源中出现的频次?
  2. 严重程度 —— 是否阻碍理解、留存或发布信任?
  3. 代表性 —— 来自目标玩家还是偏差受众?
  4. 时间风险 —— 忽略该问题是否会损害下一场活动、Demo节点或发布?
然后将每个标记为以下之一:
  • fix-before-next-build
  • watch-next-test
  • do-not-overreact-yet
单一的强烈评论并非自动成为最高优先级。

Step 4: Build the feedback brief

步骤4:生成反馈简报

Return a concise report with this exact structure:
markdown
undefined
返回结构完全一致的简洁报告:
markdown
undefined

Game Demo Feedback Brief

Game Demo Feedback Brief

Scope

Scope

  • Source mix: ...
  • Build stage: ...
  • Audience fit: ...
  • Target constraint: ...
  • Confidence: high | medium | low
  • Source mix: ...
  • Build stage: ...
  • Audience fit: ...
  • Target constraint: ...
  • Confidence: high | medium | low

Highest-confidence signals

Highest-confidence signals

  • 2-4 bullets on what the evidence most strongly suggests
  • 2-4 bullets on what the evidence most strongly suggests

Priority buckets

Priority buckets

BucketWhy it matters nowEvidence weightAction
......high / medium / lowfix-before-next-build / watch-next-test / do-not-overreact-yet
BucketWhy it matters nowEvidence weightAction
......high / medium / lowfix-before-next-build / watch-next-test / do-not-overreact-yet

Top 3 fixes before the next beat

Top 3 fixes before the next beat

  1. ...
  2. ...
  3. ...
  1. ...
  2. ...
  3. ...

Issues to watch, not overreact to

Issues to watch, not overreact to

  • ...
  • ...
  • ...
  • ...

Evidence gaps

Evidence gaps

  • ...
  • ...
  • ...
  • ...

Recommended next artifact

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
  • 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

What not to do yet

  • 1-3 bullets that prevent noisy backlog churn or premature redesigns
undefined
  • 1-3 bullets that prevent noisy backlog churn or premature redesigns
undefined

Step 5: Tailor the brief to the request type

步骤5:根据请求类型调整简报

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
针对内部Playtest:
  • 在长尾打磨前优先考虑理解度、趣味性和进度
  • 区分团队已知问题与真正的新玩家困惑
针对Steam Playtest / Next Fest / 公开Demo反馈:
  • 更重视声誉风险
  • 区分Demo质量阻塞问题与商店页面或流量问题
  • 明确下一步最佳行动是修复Demo、调整宣传信息还是决策时间节点
针对创作者/主播反馈:
  • 聚焦首次会话的清晰度、可读性、展示时刻,以及流程变得不可观看或不可分享的节点
  • 区分面向观众的宣传信息问题与深层系统平衡投诉
针对混合Bug + 性能 + UX反馈包:
  • stability-performance
    作为分类桶,而非整体结论
  • 若需要详细分析,推荐
    game-performance-profiler
    作为后续工件,再将结果整合回简报

Step 6: Ask for the minimum missing evidence when needed

步骤6:必要时请求最少的缺失证据

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.
若可信度较低,请求能实质性改善分类整理的最小补充包:
  1. 5-10条代表性反馈引用或问题要点
  2. 测试者是否为目标玩家或友好/内部测试者
  3. 当前版本阶段和下一个外部里程碑
  4. 任何崩溃/性能模式或特定设备投诉
  5. 关于玩家在何处放弃、困惑或失去兴趣的简短总结
不要要求冗长的事后分析。

Output format

输出格式

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
始终返回简短的分类整理简报,而非长篇设计文稿。
必备特质:
  • 优先处理最严重的单个玩家侧阻塞问题
  • 区分代表性模式与轶事 outliers
  • 报告字数保持在约350-550字,除非用户要求更多
  • 推荐一个后续工件,而非十个并行工作流
  • 涉及公开节点时,明确说明发布/展会时间风险

Examples

示例

Example 1: Next Fest demo confusion

示例1:Next Fest Demo困惑问题

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
输入
我们对Next Fest Demo进行了小型Playtest。大多数人认为操控尚可,但很多测试者在前十分钟就迷路了,一些主播表示不确定游戏的真正亮点是什么。还有少数人提到Steam Deck上存在卡顿问题。
输出框架
  • 主分类桶:
    onboarding-comprehension
  • 次分类桶:
    store-demo-conversion-message
    stability-performance
  • 前3项修复聚焦早期清晰度、首次会话指引,以及针对性的Steam Deck跟进,而非全面重设计
  • 推荐后续工件:
    onboarding fix brief

Example 2: Mixed internal notes before creator outreach

示例2:创作者推广前的混合内部记录

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
输入
在我们将Demo发送给创作者前,把这些Playtest记录转化为优先级事项。Bug列表很长,但真正的模式似乎是人们喜欢战斗,但在菜单环节就放弃了。
输出框架
  • 最高可信度信号:核心玩法已被接受,但菜单/操控/可读性问题影响了留存
  • 优先级行动:在打磨低频次Bug前,先修复高摩擦的菜单/HUD可读性问题
  • 推荐后续工件:
    demo polish checklist
    creator-feedback summary

Example 3: Hard feedback packet with store/demo confusion

示例3:涉及商店/Demo困惑的复杂反馈包

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
输入
我们的Steam Playtest注册量不错,但玩家要么在教程后流失,要么表示从页面和预告片里预期的是另一种游戏。我们应该先修复什么?
输出框架
  • 分类桶:
    onboarding-comprehension
    store-demo-conversion-message
  • 简报区分预期设定与游戏玩法引导
  • 推荐后续工件:
    store/demo message update
    onboarding fix brief
  • 暂不行动:不要将问题掩盖在宽泛的范围扩张下

Best practices

最佳实践

  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.
  1. 将反馈视为证据而非投票 —— 优先处理代表性阻塞问题,而非强烈的轶事。
  2. 区分理解度、趣味性与稳定性 —— 这些会产生不同的后续工件。
  3. 关注下一个外部节点 —— Next Fest、创作者推广和发布窗口会改变权重。
  4. 使用一份简短的优先级简报 —— 避免用所有问题淹没团队。
  5. 必要时转介技术深度分析 —— 详细性能分析和原始日志分类整理属于专门技能范畴。
  6. 避免积压工作混乱 —— 将一些问题标记为“观察下一次测试”而非立即修复。

References

参考资料