关于「每次点击收益:佣金率之外的四个乘数与选品前沿」,《图文截流与联盟 EPC》给出的定量结论是什么?
结论:第四章说明每次点击收益是五个乘数的乘积,示意商品里佣金率与 EPC 的排序相关为 -0.262,按佣金率挑出的第一名只拿到最优的 0.776,并给出选品前沿与时间预算下的精确解。
数据口径:已发表的研究只支持争议与讨论之间的倒 U 形,并没有测量购买;把下游转化惩罚乘上去,示意参数下讨论顶点在争议度 6.00,收入顶点在 3.50,按热闹发帖要损失 0.268。 书中参数为示意,全部计算由书中程序生成。
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。 程序只用 Python 标准库与结论吻合的打印值: -0.262, 0.776
| item | rate | cvr | aov RM | EPC RM | rank | rank |
|---|---|---|---|---|---|---|
| EPC | rate | |||||
| A kitchen scale | 0.04 | 0.060 | 89 | 0.1906 | 5 | 7 |
| B air fryer | 0.03 | 0.035 | 329 | 0.2954 | 1 | 8 |
| C skincare set | 0.12 | 0.018 | 129 | 0.2084 | 4 | 3 |
| D phone case | 0.10 | 0.045 | 29 | 0.1141 | 8 | 4 |
| E course voucher | 0.30 | 0.006 | 199 | 0.2292 | 2 | 1 |
| F luggage | 0.05 | 0.020 | 259 | 0.2142 | 3 | 6 |
| G desk lamp | 0.06 | 0.040 | 79 | 0.1693 | 6 | 5 |
| H gadget bundle | 0.20 | 0.008 | 159 | 0.1526 | 7 | 2 |
| Spearman correlation of the two rankings: -0.262 | ||||||
| top by rate earns 0.776 of the top by EPC |
查看完整验算程序
# Chapter 4. p41: earnings per click, not the commission rate.
# EPC = rate x cvr x aov x (1 - returns) x keep, where keep is the
# share of orders that stay attributed inside the programme's window.
# Eight illustrative catalogue items in RM; no real programme's rates.
import math
ITEMS = [
# name, rate, cvr, aov, ret, keep
("A kitchen scale", 0.04, 0.060, 89.0, 0.03, 0.92),
("B air fryer", 0.03, 0.035, 329.0, 0.05, 0.90),
("C skincare set", 0.12, 0.018, 129.0, 0.12, 0.85),
("D phone case", 0.10, 0.045, 29.0, 0.08, 0.95),
("E course voucher", 0.30, 0.006, 199.0, 0.20, 0.80),
("F luggage", 0.05, 0.020, 259.0, 0.06, 0.88),
("G desk lamp", 0.06, 0.040, 79.0, 0.04, 0.93),
("H gadget bundle", 0.20, 0.008, 159.0, 0.25, 0.80),
]
def epc(rate, cvr, aov, ret, keep):
return rate * cvr * aov * (1.0 - ret) * keep
def ranks(vals):
order = sorted(range(len(vals)), key=lambda i: -vals[i])
r = [0] * len(vals)
for pos, i in enumerate(order):
r[i] = pos + 1
return r
if __name__ == "__main__":
e = [epc(*it[1:]) for it in ITEMS]
rates = [it[1] for it in ITEMS]
re_, rr = ranks(e), ranks(rates)
print(" item rate cvr aov RM EPC RM rank rank")
print(" EPC rate")
for it, v, a, b in zip(ITEMS, e, re_, rr):
print(" {:<17} {:.2f} {:.3f} {:>6.0f} {:>7.4f} {:>3} {:>4}"
.format(it[0], it[1], it[2], it[3], v, a, b))
n = len(ITEMS)
d2 = [(a - b) ** 2 for a, b in zip(re_, rr)]
tot = int(math.fsum(d2))
rho = 1 - 6 * tot / (n * (n * n - 1))
print("Spearman correlation of the two rankings: {:.3f}".format(rho))
best_rate = max(range(n), key=lambda i: rates[i])
best_epc = max(range(n), key=lambda i: e[i])
print("top by rate earns {:.3f} of the top by EPC".format(
e[best_rate] / e[best_epc]))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《公域图文社交信息流截流、评论区置顶桥接与电商联盟高佣变现母典》第 4 章「每次点击收益:佣金率之外的四个乘数与选品前沿」(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-354 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026354,
author = {Fan, Yafu},
title = {Text-First Social Feed Traffic Interception, Pinned-Comment Bridge Funnels and High-Commission Affiliate Monetisation},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}