Back to skills
SKILL.md
1358 Bot Revenue Attribution Analysis 0fc6dd23
ASecurity分析各Bot的收入贡献和毛利率表现,识别高价值Bot和优化机会。
- 9 stars
- 0 votes
- 0 copies
- 1 view
- Added October 11, 2026
Works with
Security analysis
100/100npx -y skills add tools-only/X-Skills --skill 1358-bot-revenue-attribution-analysis_0fc6dd23 --agent claude-codeAre you the author of 1358 Bot Revenue Attribution Analysis 0fc6dd23?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/tools-only-1358-bot-revenue-attribution-analysis-0fc6dd23)# Bot收入归因分析模板
## 目标
分析各Bot的收入贡献和毛利率表现,识别高价值Bot和优化机会。
## 参数
- **时间范围**: 过去7天/30天
- **归因模型**: Last Touch Optimized(推荐,覆盖率70-80%)
- **最小任务数**: 10(过滤低流量Bot)
## 数据源
**PostgreSQL表**(Vercel Functions自动生成):
- `daaf_bot_revenue_snapshots` - Bot级别归因数据
- `daaf_daily_summary_snapshots` - 每日汇总数据
## Step 1: 获取Top Bot(按毛利额排序)
```sql
-- 使用Last Touch Optimized模型
WITH bot_metrics AS (
SELECT
bot_id,
bot_name,
SUM(task_count) as total_tasks,
SUM(last_touch_opt_revenue_usd) as revenue_usd,
SUM(total_cost_usd) as cost_usd,
SUM(last_touch_opt_revenue_usd) - SUM(total_cost_usd) as margin_usd,
SUM(last_touch_opt_order_count) as order_count,
CASE
WHEN SUM(last_touch_opt_revenue_usd) > 0
THEN ((SUM(last_touch_opt_revenue_usd) - SUM(total_cost_usd)) / SUM(last_touch_opt_revenue_usd)) * 100
ELSE NULL
END as margin_pct
FROM daaf_bot_revenue_snapshots
WHERE snapshot_date >= CURRENT_DATE - INTERVAL '7 days'
GROUP BY bot_id, bot_name
HAVING SUM(task_count) >= 10 -- 至少10个任务
)
SELECT
bot_id,
bot_name,
total_tasks,
ROUND(revenue_usd::numeric, 2) as revenue_usd,
ROUND(cost_usd::numeric, 2) as cost_usd,
ROUND(margin_usd::numeric, 2) as margin_usd,
ROUND(margin_pct::numeric, 2) as margin_pct,
order_count
FROM bot_metrics
ORDER BY margin_usd DESC
LIMIT 20;
```
**预期输出**:
```
| bot_id | bot_name | total_tasks | revenue | cost | margin | margin_pct | orders |
|------------|--------------------|-------------|---------|-------|--------|------------|--------|
| 1747502172 | Ultra Head Swap | 229 | 487.30 | 4.58 | 482.72 | 99.06 | 24 |
| 1764937942 | AI Nude Generator | 432 | 421.80 | 8.64 | 413.16 | 97.95 | 21 |
| 1764754691 | Face Swap Porn | 526 | 398.50 | 10.52 | 387.98 | 97.36 | 19 |
```
## Step 2: 识别负毛利Bot
```sql
-- 找出成本大于收入的Bot
SELECT
bot_id,
bot_name,
SUM(task_count) as total_tasks,
SUM(last_touch_opt_revenue_usd) as revenue_usd,
SUM(total_cost_usd) as cost_usd,
SUM(total_cost_usd) - SUM(last_touch_opt_revenue_usd) as loss_usd
FROM daaf_bot_revenue_snapshots
WHERE snapshot_date >= CURRENT_DATE - INTERVAL '7 days'
GROUP BY bot_id, bot_name
HAVING SUM(total_cost_usd) > SUM(last_touch_opt_revenue_usd)
AND SUM(task_count) >= 10
ORDER BY loss_usd DESC
LIMIT 10;
```
**行动建议**:
- 评估是否需要调整定价
- 考虑限制免费用户使用
- 优化Bot算法降低成本
## Step 3: 归因覆盖率分析
```sql
-- 查看每日归因覆盖率趋势
SELECT
snapshot_date,
order_count,
attributed_order_count,
ROUND(attribution_coverage_pct::numeric, 2) as coverage_pct
FROM daaf_daily_summary_snapshots
WHERE snapshot_date >= CURRENT_DATE - INTERVAL '30 days'
ORDER BY snapshot_date DESC;
```
**目标覆盖率**: 70-80%
如果覆盖率低于50%,说明:
1. 大量新用户未使用Bot就付费(需要优化onboarding)
2. 归因窗口太窄(考虑扩展到±14天)
3. 用户在付费后才首次使用(考虑提供试用激励)
## Step 4: 比较三种归因模型
```sql
-- 对比三种模型的差异
SELECT
bot_id,
bot_name,
ROUND(SUM(proportional_revenue_usd)::numeric, 2) as proportional_rev,
ROUND(SUM(last_touch_revenue_usd)::numeric, 2) as last_touch_rev,
ROUND(SUM(last_touch_opt_revenue_usd)::numeric, 2) as last_touch_opt_rev,
SUM(proportional_order_count) as prop_orders,
SUM(last_touch_order_count) as lt_orders,
SUM(last_touch_opt_order_count) as lto_orders
FROM daaf_bot_revenue_snapshots
WHERE snapshot_date >= CURRENT_DATE - INTERVAL '7 days'
AND bot_id IN (1747502172, 1764937942, 1764754691) -- Top 3 bots
GROUP BY bot_id, bot_name;
```
**模型选择建议**:
- **Proportional**: 用于公平分配,但可能低估明星Bot
- **Last Touch**: 简单直接,但漏掉付费后才使用的情况
- **Last Touch Optimized**: 最推荐,覆盖率最高
## Step 5: 生成可视化报告
### 5.1 Top 10 Bot毛利额条形图
```javascript
// 数据转换(从Step 1查询结果)
const data = [
{ category: 'Ultra Head Swap', value: 482.72 },
{ category: 'AI Nude Generator', value: 413.16 },
{ category: 'Face Swap Porn', value: 387.98 },
// ... 其他7个
];
```
使用MCP Chart工具生成条形图:
```
mcp__mcphub__mcp-server-chart-generate_bar_chart
```
### 5.2 每日归因覆盖率趋势线图
```javascript
// 数据转换(从Step 3查询结果)
const data = [
{ time: '2025-12-29', value: 72.5, group: '覆盖率%' },
{ time: '2025-12-30', value: 68.3, group: '覆盖率%' },
// ...
];
```
使用MCP Chart工具生成折线图。
## 关键指标
### 整体指标
- **总收入归因**: 所有Bot的LTO模型收入总和
- **归因覆盖率**: 有任务历史的订单 / 总订单
- **平均每订单归因收入**: 总归因收入 / 归因订单数
### Bot级别指标
- **毛利额 (Margin USD)**: Revenue - Cost
- **毛利率 (Margin %)**: (Revenue - Cost) / Revenue × 100%
- **每任务平均收入**: Revenue / Task Count
- **每订单平均任务数**: Task Count / Order Count
## 优化建议
### 高毛利Bot(>90%)
✅ **策略**: 加大推广力度
- 增加首页展示位
- 社交媒体营销
- 优化SEO
### 中等毛利Bot(50-90%)
⚠️ **策略**: 平衡增长和成本
- 监控成本趋势
- 考虑分级定价
- 优化算法效率
### 低/负毛利Bot(<50%)
❌ **策略**: 紧急优化
- 限制免费用户额度
- 提高付费转化率
- 降低算法成本
- 必要时下线Bot
## 自动化报告
可以创建定时任务每周生成报告:
```typescript
// functions/api/cron/weekly-bot-report.ts
// 每周一02:00 UTC发送Slack报告
```
包含:
1. Top 10高毛利Bot
2. 需要关注的负毛利Bot
3. 归因覆盖率趋势
4. 周环比变化
## 注意事项
⚠️ **收入与成本日期不匹配**
- 收入归因到订单日期
- 成本记录在任务执行日期
- 单日数据可能有波动,建议看7天或30天趋势
⚠️ **新Bot冷启动**
- 前7-14天可能无归因数据
- 需要积累足够订单历史
- 初期重点关注成本效率
⚠️ **季节性影响**
- 节假日订单量波动
- 考虑同比而非环比
- 识别周期性模式
## 下一步行动
1. ✅ 部署归因系统到Vercel
2. ✅ 等待24小时收集首批数据
3. ✅ 运行本分析模板
4. ✅ 识别Top 5高价值Bot和Top 5需优化Bot
5. ✅ 制定具体的增长/优化策略
6. ✅ 设置每周自动报告
Attribution
Comments
Loading comments…
Newsletter
Get the weekly list of safe new skills, and the ones we pulled.