关于「退订与投诉的红线:停发规则、投诉刹车与三地同意法规」,《弃单召回与升单棘轮》给出的定量结论是什么?

结论:真实投诉率为 0.0006 时,每周 1,000 封有 0.451 的星期碰线,7,000 封时碰刹车线的机会是 0.247,碰 0.10 % 线的机会 0.132;35,000 封时分别是 0.082 与 0.003。真实投诉率为 0.0012 时,名单本身已经越过了 0.10 % 的线:每周 7,000 封时,0.843 的星期会碰刹车线,0.733 的星期碰 0.10 % 的线;35,000 封时分别是 0.991 与 0.879。这时刹车不是误报,而是如实报告。

数据口径:按本书的示意模型,第一波召回放在一分钟时只挽回同意离开者的 0.00609,放在十五分钟时挽回 0.03732;把三波压在一小时之内,每位离开者净值 RM 1.1790,按十五分钟、二十四小时、七十二小时发送时为 RM 1.9419;第五波的净值已是 -0.0121。 书中网站被当作付费报告产品讨论,不宣称任何预测准确性,不用恐惧招揽;所有访客与订单数据均为合成或示意。

English version

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: 0.0006, 0.0012, 0.003, 0.082, 0.10, 0.132

A. chance that one week's observed rate reaches a line, by weekly sends
true rate sends/week >= brake 0.08 % >0.10 % >= 0.30 %
0.00031,000 0.259 0.259 0.004
0.00037,000 0.020 0.006 0.000
0.000335,000 0.000 0.000 0.000
0.00061,000 0.451 0.451 0.023
0.00067,000 0.247 0.132 0.000
0.000635,000 0.082 0.003 0.000
0.00121,000 0.699 0.699 0.120
0.00127,000 0.843 0.733 0.000
0.001235,000 0.991 0.879 0.000
B. zero complaints observed: one-sided 95 % upper bound 1 - 0.05^(1/n)
n300: bound 0.00994
n1,000: bound 0.00299
n3,000: bound 0.00100
n10,000: bound 0.00030
sends needed before a clean record shows the rate is below 0.10 %2,995
查看完整验算程序
# Chapter 4. p42: complaint-rate brake against the published Gmail lines.
# Google's sender guidelines ask bulk senders to keep the spam rate in
# Postmaster Tools below 0.10 % and never to reach 0.30 %. Internal brake
# (this book's rule): pause all recall waves when the 7-day complaint
# rate reaches 0.08 %. Binomial arithmetic, exact tails; send volumes
# and true rates are illustrative.
import math

LINE_SOFT, LINE_HARD, BRAKE = 0.0010, 0.0030, 0.0008


def tail_ge(n, p, k):
    # P(X >= k) for X ~ Binomial(n, p), summed in log space
    if k <= 0:
        return 1.0
    terms = []
    for x in range(k, n + 1):
        lt = (math.lgamma(n + 1) - math.lgamma(x + 1) - math.lgamma(n - x + 1)
              + x * math.log(p) + (n - x) * math.log1p(-p))
        if lt < -60 and x > n * p:
            break
        terms.append(math.exp(lt))
    return math.fsum(terms)


if __name__ == "__main__":
    print("A. chance that one week's observed rate reaches a line, by weekly sends")
    print("   true rate   sends/week   >= brake 0.08 %   >= 0.10 %   >= 0.30 %")
    for p in (0.0003, 0.0006, 0.0012):
        for n in (1000, 7000, 35000):
            kb = math.ceil(BRAKE * n)
            ks = math.ceil(LINE_SOFT * n)
            kh = math.ceil(LINE_HARD * n)
            print("   {:.4f}      {:>6,}       {:.3f}             {:.3f}       {:.3f}".format(
                p, n, tail_ge(n, p, kb), tail_ge(n, p, ks), tail_ge(n, p, kh)))
    print("B. zero complaints observed: one-sided 95 % upper bound 1 - 0.05^(1/n)")
    for n in (300, 1000, 3000, 10000):
        print("   n = {:>6,}: bound {:.5f}".format(n, 1.0 - 0.05 ** (1.0 / n)))
    n = 1
    while 1.0 - 0.05 ** (1.0 / n) >= LINE_SOFT:
        n = n + 1
    print("   sends needed before a clean record shows the rate is below 0.10 %: {:,}".format(n))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《诊断收费站弃单挽回时序、个性化悬念召回与高客单升单棘轮母典》第 4 章「退订与投诉的红线:停发规则、投诉刹车与三地同意法规」(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-349 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026349,
  author    = {Fan, Yafu},
  title     = {Paywall Abandonment Recovery Timing, Personalised Curiosity Recall and High-Ticket Upsell Ratchets},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论