关于「试点定金:分离两类品牌方」,《广告库反查与进箱物理》给出的定量结论是什么?
结论:第六章用信号与筛选理论分析对称试点定金,示意参数下的分离线约为 1,022.73,认真品牌方接受比例须高于 0.585 才不减少总收入。
数据口径:公开广告库里仍在投的广告是右删失数据,须用卡普兰与迈耶的乘积极限估计去读;「仍在投且已投满 30 天」这条规则在合成广告库里把持续投放者的比例从 0.313 抬到 0.902。 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: 1022.73
| separation line D* = w / ((1 - e) - e m) = 11,250/11 = 1,022.73 | ||||
| D | F payoff | joins? | S payoff | agency if F joins |
| 0 | 900.00 | yes | 4,000.00 | -1,200.00 |
| 500 | 460.00 | yes | 4,010.00 | -760.00 |
| 1,000 | 20.00 | yes | 4,020.00 | -320.00 |
| 1,023 | -0.24 | no | 4,020.46 | -299.76 |
| 1,500 | -420.00 | no | 4,030.00 | 120.00 |
| 2,000 | -860.00 | no | 4,040.00 | 560.00 |
| expected late credit to the brand at D = 1,500: 30 (30.00) | ||||
| a brand that walks away pays at most D; an agency that is late | ||||
| pays (1 + m) D: the terms bind both sides |
查看完整验算程序
# Chapter 6. p61: a refundable pilot deposit as a screening device.
# Exact arithmetic with Fraction. Illustrative amounts in RM.
# Two kinds of brand ask for a custom pilot proposal:
# serious (S): would run the pilot; net value to them b_S = 4,000
# free-rider (F): wants only the proposal, worth w = 900 to them
# Symmetric terms: the deposit D is credited to the pilot fee if the
# pilot runs; refunded in full plus a late credit m * D if the agency
# misses the written proposal or kickoff date (chance e); kept if the
# brand walks away after receiving the proposal and the kickoff.
# Agency cost of a custom proposal and kickoff c = 1,200.
from fractions import Fraction as F
B_S, W, C = F(4000), F(900), F(1200)
E, M = F(1, 10), F(1, 5)
def free_rider(d):
return W - (1 - E) * d + E * M * d
def serious(d):
return B_S + E * M * d
def agency_on_free_rider(d):
return (1 - E) * d - E * M * d - C
if __name__ == "__main__":
dstar = W / ((1 - E) - E * M)
print("separation line D* = w / ((1 - e) - e m) = {:,}/{} = {:,.2f}"
.format(dstar.numerator, dstar.denominator, float(dstar)))
print(" D F payoff joins? S payoff agency if F joins")
for d in (F(0), F(500), F(1000), F(1023), F(1500), F(2000)):
fp = free_rider(d)
print(" {:>5,} {:>9,.2f} {:<6} {:>9,.2f} {:>12,.2f}".format(
int(d), float(fp), "yes" if fp >= 0 else "no",
float(serious(d)), float(agency_on_free_rider(d))))
d = F(1500)
gain_brand = E * M * d
print("expected late credit to the brand at D = 1,500: {} ({:.2f})"
.format(gain_brand, float(gain_brand)))
print("a brand that walks away pays at most D; an agency that is late")
print("pays (1 + m) D: the terms bind both sides")唯一数据源与引用
范雅福(Fan, Y.). (2026). 《公开广告库金主反查、B2B冷触达进箱物理与高客单暗投签单母典》第 6 章「试点定金:分离两类品牌方」(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-347 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026347,
author = {Fan, Yafu},
title = {Active Ad-Library Prospecting, B2B Outbound Deliverability Physics and High-Ticket Whitelist Deal Closing},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}