关于「完播、互动、复播:三把尺子的统计性质」,《短视频流量池冷启动》给出的定量结论是什么?

结论:第四章说明 7 秒视频的完播率是 60 秒视频的 2.89 倍,完全来自长度,区分互动率 0.03 与 0.04 需要 5,301 次曝光。

数据口径:书中所有门槛、池子规模与权重均为示意,平台规则以其官方公示为准;全部计算由书中程序生成;本书是非个人化经营建议,只讨论真实内容质量与合规测试,不涉及任何刷量或规避平台规则的做法。 第三章说明硬门槛在数据上是斜坡,批次从 100 人增至 3,000 人时放行区间宽度从 0.124 收窄到 0.023;示意基准视频的破圈概率为 0.03585,瓶颈在第四道门。

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: 0.03, 0.04, 5301, 60

A. interaction rate 0.04 seen on n views, Wilson 95% interval
n300 0.0230 to 0.0686 width 0.0456
n3000 0.0336 to 0.0476 width 0.0141
n30000 0.0378 to 0.0423 width 0.0044
views per video to tell 0.03 from 0.045301
B. replays per view, same loop habit rho, completion from p41
L(s) completion rho0.2 rho=0.3 per completer at 0.3
70.600 0.150 0.257 0.429
150.511 0.128 0.219 0.429
300.379 0.095 0.162 0.429
600.208 0.052 0.089 0.429

核心公式

\[C(L) = (1-h)\,e^{-\lambda (L-3)},\]
查看完整验算程序
# Chapter 4. p42: how much an early batch can say about engagement,
# and why replay rewards short videos mechanically.
# Part A: Wilson 95% interval (Wilson 1927) for an interaction rate
# measured on n views, and the views needed per video to tell 0.03
# from 0.04 with two-sided alpha 0.05 and power 0.80 (normal approx).
# Part B: a viewer who finishes the video lets it loop again with
# probability rho; expected replays per view = C * rho / (1 - rho),
# where C is the completion rate from p41. Values are illustrative.
import math

Z = 1.959964
ZB = 0.841621


def wilson(x, n):
    p = x / n
    den = 1.0 + Z * Z / n
    mid = (p + Z * Z / (2.0 * n)) / den
    half = Z * math.sqrt(p * (1.0 - p) / n + Z * Z / (4.0 * n * n)) / den
    return mid - half, mid + half


print("A. interaction rate 0.04 seen on n views, Wilson 95% interval")
for n in (300, 3000, 30000):
    lo, hi = wilson(0.04 * n, n)
    print("   n = %6d   %.4f to %.4f   width %.4f" % (n, lo, hi, hi - lo))
p1, p2 = 0.03, 0.04
pb = (p1 + p2) / 2.0
need = (Z * math.sqrt(2 * pb * (1 - pb))
        + ZB * math.sqrt(p1 * (1 - p1) + p2 * (1 - p2))) ** 2 \
    / (p2 - p1) ** 2
print("   views per video to tell 0.03 from 0.04: %d" % math.ceil(need))
print("B. replays per view, same loop habit rho, completion from p41")
print("   L(s)  completion  rho=0.2  rho=0.3  per completer at 0.3")
for L, C in ((7, 0.600), (15, 0.511), (30, 0.379), (60, 0.208)):
    r2 = C * 0.2 / 0.8
    r3 = C * 0.3 / 0.7
    print("   %4d  %10.3f  %7.3f  %7.3f  %8.3f" % (L, C, r2, r3, r3 / C))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《全域短视频赛马池冷启动动力学、完播互动复播阶跃阈值与零投流破圈测试矩阵母典》第 4 章「完播、互动、复播:三把尺子的统计性质」(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-273 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026273,
  author    = {Fan, Yafu},
  title     = {Cold-Start Dynamics of Short-Video Traffic Pools: Step Thresholds in Completion, Engagement and Replay and a Zero-Ad-Spend Breakout Test Matrix},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论