自动结论
5
Evidence-backed insights
自动分析结论
本机确定性规则始终先计算事实；可选 AI 只在手动触发时解释脱敏汇总。每条结论同时展示计算依据、样本量和原始测试记录，避免凭感觉判断。
优先处理
2
有数据对象
3/4
精确数据点
43
需要关注
规则生成已有 3/4 个对象返回测试数据
广州科鑫 尚无测试批次，当前横向判断只覆盖已有数据对象。
计算依据
覆盖率 = 3 ÷ 4 = 75.0%；共 4 个批次、43 个精确数据点。
建议动作
优先确认无数据对象的测试状态和预计回数时间，再做完整排名。
证据：play800证据：广州池骋证据：广州科鑫证据：广西天将
优先处理
规则生成成熟留存证据不足
广西天将 已有测试批次，但没有精确 D7 留存。
计算依据
1/3 个有测试批次的对象缺少精确 D7；区间、上下限和缺失值不计入样本。
建议动作
向对应发行商补要成熟 D7，并确认统计口径、样本截止日和版本。
证据：广西天将
事实信号
规则生成当前精确样本中的指标领先对象
最低平均 CPI：广州池骋；最高平均 D1：play800
计算依据
CPI 均值基于 n=4 个精确数据点、2 个批次；D1 均值基于 n=1 个精确数据点、1 个批次
建议动作
把领先结果作为复测假设，不直接等同于发行商或版本胜负；先核对国家、素材、版本和日期口径。
证据：广州池骋证据：play800
需要关注
规则生成不同对象 D1 均值相差 8.06 个百分点
同一游戏在不同测试来源下的 D1 表现存在差异，需要先排除投放口径和样本结构影响。
计算依据
最高 13.84% − 最低 5.78% = 8.06 个百分点；涉及 4 个精确数据点。
建议动作
统一国家、平台、版本、素材和统计窗口后安排复测，避免把渠道差异误判为产品变化。
证据：play800证据：广州池骋
优先处理
规则生成当前建议：继续优化并补齐数据后复测
现有数据可以定位问题和提出复测假设，但不足以支持完整发行商排名或扩大投放。
计算依据
决策规则：无数据对象 1 个；有批次但缺精确 D7 的对象 1 个；报告质量风险 3 项。任一关键缺口存在时不输出“扩大投放”结论。
建议动作
先补齐无数据对象的测试结果和成熟 D7，再按同国家、同平台、同版本口径复测。
证据：广州科鑫证据：广西天将
OpenAI 辅助解读
未配置，不会发送数据只发送匿名对象、汇总指标、样本数和本机规则结论。不会发送游戏名、发行商名、批次名、联系人、链接、原始文件、聊天内容或飞书凭据；请求使用 store: false。
开关：关闭API Key：未配置模型：未配置
查看实际允许外发的脱敏载荷
SHA-256：47ebafb576d3ba0c30e9070fae39f14f9c10cd678c2cb4f0a631a6efa5b97f87
{
"schemaVersion": 1,
"reportMode": "publisher",
"totals": {
"objectCount": 4,
"batchCount": 4,
"exactPointCount": 43,
"riskCount": 3
},
"objects": [
{
"ref": "object:1",
"alias": "对象A",
"batchCount": 1,
"exactPointCount": 6,
"metrics": [
{
"ref": "object:1:metric:cpi",
"code": "cpi",
"valueType": "CURRENCY",
"average": 1.22,
"minimum": 1.22,
"maximum": 1.22,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": "USD",
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:1:metric:ctr",
"code": "ctr",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:1:metric:ipm",
"code": "ipm",
"valueType": "NUMBER",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:1:metric:retention_d1",
"code": "retention_d1",
"valueType": "PERCENT",
"average": 0.138381,
"minimum": 0.138381,
"maximum": 0.138381,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": null,
"mixedCurrency": false,
"isBest": true
},
{
"ref": "object:1:metric:retention_d7",
"code": "retention_d7",
"valueType": "PERCENT",
"average": 0.013888,
"minimum": 0.013888,
"maximum": 0.013888,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": null,
"mixedCurrency": false,
"isBest": true
},
{
"ref": "object:1:metric:roas_d1",
"code": "roas_d1",
"valueType": "PERCENT",
"average": 0.276113,
"minimum": 0.276113,
"maximum": 0.276113,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": null,
"mixedCurrency": false,
"isBest": true
},
{
"ref": "object:1:metric:roas_d7",
"code": "roas_d7",
"valueType": "PERCENT",
"average": 0.628745,
"minimum": 0.628745,
"maximum": 0.628745,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": null,
"mixedCurrency": false,
"isBest": true
},
{
"ref": "object:1:metric:ltv",
"code": "ltv",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
}
]
},
{
"ref": "object:2",
"alias": "对象B",
"batchCount": 2,
"exactPointCount": 37,
"metrics": [
{
"ref": "object:2:metric:cpi",
"code": "cpi",
"valueType": "CURRENCY",
"average": 0.5675,
"minimum": 0.33,
"maximum": 0.75,
"sampleCount": 4,
"batchCount": 2,
"currencyCode": "CNY",
"mixedCurrency": false,
"isBest": true
},
{
"ref": "object:2:metric:ctr",
"code": "ctr",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:ipm",
"code": "ipm",
"valueType": "NUMBER",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:retention_d1",
"code": "retention_d1",
"valueType": "PERCENT",
"average": 0.0578,
"minimum": 0.0454,
"maximum": 0.0706,
"sampleCount": 3,
"batchCount": 2,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:retention_d7",
"code": "retention_d7",
"valueType": "PERCENT",
"average": 0.0073,
"minimum": 0.0073,
"maximum": 0.0073,
"sampleCount": 1,
"batchCount": 1,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:roas_d1",
"code": "roas_d1",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:roas_d7",
"code": "roas_d7",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:2:metric:ltv",
"code": "ltv",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
}
]
},
{
"ref": "object:3",
"alias": "对象C",
"batchCount": 0,
"exactPointCount": 0,
"metrics": [
{
"ref": "object:3:metric:cpi",
"code": "cpi",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:ctr",
"code": "ctr",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:ipm",
"code": "ipm",
"valueType": "NUMBER",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:retention_d1",
"code": "retention_d1",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:retention_d7",
"code": "retention_d7",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:roas_d1",
"code": "roas_d1",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:roas_d7",
"code": "roas_d7",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:3:metric:ltv",
"code": "ltv",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
}
]
},
{
"ref": "object:4",
"alias": "对象D",
"batchCount": 1,
"exactPointCount": 0,
"metrics": [
{
"ref": "object:4:metric:cpi",
"code": "cpi",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:ctr",
"code": "ctr",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:ipm",
"code": "ipm",
"valueType": "NUMBER",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:retention_d1",
"code": "retention_d1",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:retention_d7",
"code": "retention_d7",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:roas_d1",
"code": "roas_d1",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:roas_d7",
"code": "roas_d7",
"valueType": "PERCENT",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
},
{
"ref": "object:4:metric:ltv",
"code": "ltv",
"valueType": "CURRENCY",
"average": null,
"minimum": null,
"maximum": null,
"sampleCount": 0,
"batchCount": 0,
"currencyCode": null,
"mixedCurrency": false,
"isBest": false
}
]
}
],
"ruleInsights": [
{
"ref": "rule:publisher-coverage",
"level": "MEDIUM",
"title": "已有 3/4 个对象返回测试数据",
"statement": "对象C 尚无测试批次，当前横向判断只覆盖已有数据对象。",
"calculation": "覆盖率 = 3 ÷ 4 = 75.0%；共 4 个批次、43 个精确数据点。",
"recommendation": "优先确认无数据对象的测试状态和预计回数时间，再做完整排名。",
"evidenceRefs": [
"object:1",
"object:2",
"object:3",
"object:4"
]
},
{
"ref": "rule:mature-retention",
"level": "HIGH",
"title": "成熟留存证据不足",
"statement": "对象D 已有测试批次，但没有精确 D7 留存。",
"calculation": "1/3 个有测试批次的对象缺少精确 D7；区间、上下限和缺失值不计入样本。",
"recommendation": "向对应发行商补要成熟 D7，并确认统计口径、样本截止日和版本。",
"evidenceRefs": [
"object:4"
]
},
{
"ref": "rule:current-leaders",
"level": "INFO",
"title": "当前精确样本中的指标领先对象",
"statement": "最低平均 CPI：对象B；最高平均 D1：对象A",
"calculation": "CPI 均值基于 n=4 个精确数据点、2 个批次；D1 均值基于 n=1 个精确数据点、1 个批次",
"recommendation": "把领先结果作为复测假设，不直接等同于发行商或版本胜负；先核对国家、素材、版本和日期口径。",
"evidenceRefs": [
"object:2",
"object:1"
]
},
{
"ref": "rule:d1-spread",
"level": "MEDIUM",
"title": "不同对象 D1 均值相差 8.06 个百分点",
"statement": "同一游戏在不同测试来源下的 D1 表现存在差异，需要先排除投放口径和样本结构影响。",
"calculation": "最高 13.84% − 最低 5.78% = 8.06 个百分点；涉及 4 个精确数据点。",
"recommendation": "统一国家、平台、版本、素材和统计窗口后安排复测，避免把渠道差异误判为产品变化。",
"evidenceRefs": [
"object:1",
"object:2"
]
},
{
"ref": "rule:decision",
"level": "HIGH",
"title": "当前建议：继续优化并补齐数据后复测",
"statement": "现有数据可以定位问题和提出复测假设，但不足以支持完整发行商排名或扩大投放。",
"calculation": "决策规则：无数据对象 1 个；有批次但缺精确 D7 的对象 1 个；报告质量风险 3 项。任一关键缺口存在时不输出“扩大投放”结论。",
"recommendation": "先补齐无数据对象的测试结果和成熟 D7，再按同国家、同平台、同版本口径复测。",
"evidenceRefs": [
"object:3",
"object:4"
]
}
],
"methodology": [
"仅统计限定方式为“精确值”的数值。",
"平均值按数据点做算术平均，不按 Spend 或 Installs 加权。",
"区间、上下限、约等于和缺失项不进入均值。",
"“最优”只依据指标方向与当前算术均值，不代表商务合作建议。"
]
}计算证据表
均值只使用精确数值；每个单元格同时显示数据点数和实际测试批次数。
| 对象 | 测试批次 | CPI | D1 留存 | D7 留存 | 原始记录 |
|---|---|---|---|---|---|
| play800狼人卡牌 | 1 | $1.22 n=1 · 1批次 | 13.84% n=1 · 1批次 | 1.39% n=1 · 1批次 | |
| 广州池骋狼人卡牌 | 2 | ¥0.57 n=4 · 2批次 | 5.78% n=3 · 2批次 | 0.73% n=1 · 1批次 | |
| 广州科鑫狼人卡牌 | 0 | — n=0 · 0批次 | — n=0 · 0批次 | — n=0 · 0批次 | 尚无数据 |
| 广西天将狼人卡牌 | 1 | — n=0 · 0批次 | — n=0 · 0批次 | — n=0 · 0批次 |
当前限制与风险
- 广州科鑫 尚无测试批次，不能进行数据判断。
- 广州科鑫、广西天将 缺少精确 D7 留存。
- 当前批次包含 7 类数据质量标记，需在批次详情核对口径。
固定计算口径
- 仅统计限定方式为“精确值”的数值。
- 平均值按数据点做算术平均，不按 Spend 或 Installs 加权。
- 区间、上下限、约等于和缺失项不进入均值。
- “最优”只依据指标方向与当前算术均值，不代表商务合作建议。
接入外部 AI 后，AI 只能解释这些已计算事实，不能改写原始数据、样本数或计算结果。