关于「周度台账与证伪闭环」,《广告库反查与进箱物理》给出的定量结论是什么?

结论:第八章给出周度台账、蒸馏小工具「金主反查评分与冷触达进箱预算器」与四项预注册检验。本书只写合规做法:公开广告库按平台条款查看,冷触达只发公开的企业联系人并如实标明身份、提供有效退订,授权投放一律标明付费合作,不写任何规避过滤的手法。书中以标注框区分已发表的证据、可运行的 Python 计算(20 段程序)、原典与作者的假说。

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。

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

A. brands from public ad-library observations
branddaysrefreshscoretier
Demo Skincare6456.67A
Demo Snacks4124.74B
Demo Fitness912.80C
Demo Travel11204.73B
Demo Fintech000.00C
Demo Fashion2366.18A
B. sending subdomain, last 14 days
today 40 compl 0 / 420 P>0.1% 0.242 P>0.3% 0.014 grow -> 50
today 80 compl 1 / 900 P>0.1% 0.434 P>0.3% 0.022 grow -> 100
today 120compl 4 / 1,500P>0.1% 0.891P>0.3% 0.132hold -> 120
today 150compl 6 / 1,800P>0.1% 0.976P>0.3% 0.267halve -> 75
查看完整验算程序
# Chapter 8. p81: one week of the prospecting and outbound ledger.
# Synthetic rows, invented brand names. For each brand: days its oldest
# live ad has run, creatives refreshed in 30 days. For the sending
# subdomain: inboxed messages and complaints over the last 14 days and
# today's volume. The functions are the ones in distill/tool.py.
import math

POOL = 150


def score(days, refresh):
    return math.log1p(days) + 0.5 * refresh


def tier(s):
    return "A" if s >= 5.5 else ("B" if s >= 4.0 else "C")


def p_exceed(k, n, x):
    big = n + 999
    terms = [math.exp(math.lgamma(big + 1) - math.lgamma(j + 1)
                      - math.lgamma(big - j + 1) + j * math.log(x)
                      + (big - j) * math.log1p(-x)) for j in range(k + 1)]
    return min(1.0, math.fsum(terms))


def next_volume(v, k, n, pool=POOL):
    hi, lo = p_exceed(k, n, 0.003), p_exceed(k, n, 0.001)
    if hi > 0.2:
        return max(10, int(v / 2)), "halve", lo, hi
    if lo < 0.5:
        return min(pool, int(v * 1.25)), "grow", lo, hi
    return v, "hold", lo, hi


BRANDS = (("Demo Skincare", 64, 5), ("Demo Snacks", 41, 2),
          ("Demo Fitness", 9, 1), ("Demo Travel", 112, 0),
          ("Demo Fintech", 0, 0), ("Demo Fashion", 23, 6))


if __name__ == "__main__":
    print("A. brands from public ad-library observations")
    print("  brand            days  refresh  score  tier")
    for name, d, r in BRANDS:
        s = score(d, r)
        print("  {:<15} {:>5}  {:>7}  {:>5.2f}  {:>4}".format(
            name, d, r, s, tier(s)))
    print("B. sending subdomain, last 14 days")
    for v, k, n in ((40, 0, 420), (80, 1, 900), (120, 4, 1500),
                    (150, 6, 1800)):
        nv, act, lo, hi = next_volume(v, k, n)
        print("  today {:>3}  compl {} / {:>5,}  P>0.1% {:.3f}  P>0.3% {:.3f}"
              "  {} -> {}".format(v, k, n, lo, hi, act, nv))

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 8 章「周度台账与证伪闭环」(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}
}

本书其他章节结论