关于「双轨合同:固定授权费、分档抽佣与委托代理的算术」,《暗投、归因与双轨抽佣》给出的定量结论是什么?
结论:第六章设计代理商与品牌之间的双轨合同:固定月费加按实验测得的增量净成交额边际分档的抽佣,并设退货追回与约 0.13 的留存比例;线性合同的最优抽佣比例随测量噪声下降,只看后台时是 0.357,地区实验时是 0.762。双轨抽佣在本书中专指单层企业间服务费,不涉及任何多层计佣。代理业务每天 RM 5,000 的服务费收入只作为推算条件的目标,不是承诺。
数据口径:在一个所有真相都已知的合成周里,三个后台认领的订单合计是实际订单的 1.112 倍,真正由广告带来的增量只占实际订单的 0.312;后台 ROAS 之和是 4.81,真实增量 ROAS 只有 1.35,低于盈亏平衡线 2.884。 核对原文之后,本书做了蓝图修正:付费向自然流量的外溢在随机实验里有正有负,本书把它写成可正可负的净效应;沙普利值适合分功劳,不适合定预算;用户级留出组对多数东南亚品牌太大,重点改为全马十六个地区的地区实验;平台费率无法逐项核实,全部作为示意输入;马来西亚电子交易条例已由 2024 年新条例取代。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 与结论吻合的打印值: 0.357, 0.762
| measurement | s | beta* | effort | risk pr | surplus | alpha | GMV rate |
|---|---|---|---|---|---|---|---|
| dashboards | 60 | 0.357 | 35.7 | 11.48 | 17.86 | 45.10 | 0.0426 |
| geo test | 25 | 0.762 | 76.2 | 9.07 | 38.10 | 20.05 | 0.0910 |
| holdout + ghost ads | 15 | 0.899 | 89.9 | 4.54 | 44.94 | 4.15 | 0.1073 |
| perfect | 0 | 1.000 | 100.0 | 0.00 | 50.00 | -10.00 | 0.1194 |
| first-best surplus 50.00; GMV rate = beta* x q, the equivalent commission on incremental GMV | |||||||
| risk aversion 0.02: beta* on dashboards 0.581, on a geo test 0.889 | |||||||
| risk aversion 0.10: beta* on dashboards 0.217, on a geo test 0.615 |
核心公式
\[\beta^* = \frac{1}{1 + r\,c\,s^2}.\]
查看完整验算程序
# Chapter 6. p63: the Holmstrom-Milgrom linear contract applied to an agency (illustrative units).
# Measured output x = e + noise, noise ~ N(0, s^2), in RM thousand of monthly incremental
# contribution. Effort cost (c/2) e^2, agency CARA risk aversion r. Pay = alpha + beta * x.
# Optimal beta* = 1 / (1 + r c s^2); effort e = beta / c. The noise s depends on HOW output is
# measured: platform dashboards, a geo test, or a user-level holdout with ghost-ad controls.
C, R, U0, Q = 0.01, 0.05, 40.0, 0.1194
print("{:<22}{:>6}{:>8}{:>8}{:>9}{:>9}{:>9}{:>10}".format(
"measurement", "s", "beta*", "effort", "risk pr", "surplus", "alpha", "GMV rate"))
first_best = 1.0 / C - 0.5 / C
for name, s in (("dashboards", 60.0), ("geo test", 25.0), ("holdout + ghost ads", 15.0), ("perfect", 0.0)):
beta = 1.0 / (1.0 + R * C * s * s)
e = beta / C
risk = 0.5 * R * beta * beta * s * s
surplus = e - 0.5 * C * e * e - risk
alpha = U0 + 0.5 * C * e * e + risk - beta * e
print("{:<22}{:>6.0f}{:>8.3f}{:>8.1f}{:>9.2f}{:>9.2f}{:>9.2f}{:>10.4f}".format(
name, s, beta, e, risk, surplus, alpha, beta * Q))
print("first-best surplus {:.2f}; GMV rate = beta* x q, the equivalent commission on incremental GMV".format(
first_best))
for r in (0.02, 0.10):
b = 1.0 / (1.0 + r * C * 60.0 * 60.0)
print("risk aversion {:.2f}: beta* on dashboards {:.3f}, on a geo test {:.3f}".format(
r, b, 1.0 / (1.0 + r * C * 25.0 * 25.0)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《东南亚社交电商暗投带货联动、站内外归因叠加与双轨抽佣增长母典》第 6 章「双轨合同:固定授权费、分档抽佣与委托代理的算术」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-351 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026351,
author = {Fan, Yafu},
title = {Southeast Asian Social-Commerce Whitelist Amplification, Cross-Channel Attribution and Dual-Track Commission Growth},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}