What does Short-Video Traffic Pools conclude, in numbers, about “The Zero-Ad-Spend Test Matrix: a Twelve-Run Plackett–Burman Design and Its Sample Size”?

Answer: Chapter 6 presents a zero-ad-spend Plackett-Burman matrix of twelve videos a week and three factors, whose main-effect variance is 0.333 of a single video's; detecting a completion difference of 0.04 takes 11 weeks.

Scope of the numbers: Chapter 3 shows that a hard threshold is a ramp in the data, whose width narrows from 0.124 to 0.023 as the batch grows from 100 to 3,000 viewers; the illustrative base video breaks out with probability 0.03585, with the bottleneck at the fourth gate. Marked boxes separate published evidence, runnable Python calculations (18 programs), classical sources and the author's hypotheses.

中文版

Data table

The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. printed values that match the answer: 0.333, 11

video A hook B length C topic
1 result40s how-to
2 question40s story
3 result15s story
4 question40s how-to
5 question15s story
6 question15s how-to
7 result15s how-to
8 result40s how-to
9 result40s story
10 question40s story
11 result15s story
12 question15s how-to
factor A: +1 in 6 videos, -1 in 6
factor B: +1 in 6 videos, -1 in 6
factor C: +1 in 6 videos, -1 in 6
columns A.B dot product0
columns A.C dot product0
columns B.C dot product0
interaction alias: AB on C -0.333, AC on B -0.333, BC on A -0.333
variance of a main effect4/12 sigma^2 = 0.333 sigma^2
one factor at a time, 3 baseline + 3 per factor:
variance1/3 + 1/3 = 0.667 sigma^2, twice the matrix
Show the full program
# Chapter 6. p61: the twelve-run Plackett-Burman matrix (1946).
# Twelve videos a week, three two-level factors (illustrative):
#   A hook   : question opening (-1) or result-first opening (+1)
#   B length : about 15 s (-1) or about 40 s (+1)
#   C topic  : how-to (-1) or behind-the-scenes story (+1)
# Rows are the cyclic shifts of the Plackett-Burman generator plus a
# row of all -1; we keep the first three columns. Every factor is +1
# in six videos and -1 in six, and every pair of columns is
# orthogonal, so each main effect is estimated from all twelve videos
# at once: effect = mean(+1) - mean(-1), variance 4 sigma^2 / 12.
# The price: each two-factor interaction leaks into the third factor's
# main effect with weight -1/3 (partial aliasing, printed below).
GEN = [1, 1, -1, 1, 1, 1, -1, -1, -1, 1, -1]
rows = [GEN[-i:] + GEN[:-i] for i in range(11)] + [[-1] * 11]
X = [r[:3] for r in rows]
label = {("A", -1): "question", ("A", 1): "result",
         ("B", -1): "15s", ("B", 1): "40s",
         ("C", -1): "how-to", ("C", 1): "story"}
print("video   A hook    B length  C topic")
for i, r in enumerate(X, 1):
    print("%5d   %-8s  %-8s  %s" % (i, label[("A", r[0])],
                                   label[("B", r[1])],
                                   label[("C", r[2])]))
cols = list(zip(*X))
for j, name in enumerate("ABC"):
    print("factor %s: +1 in %d videos, -1 in %d"
          % (name, cols[j].count(1), cols[j].count(-1)))
for a in range(3):
    for b in range(a + 1, 3):
        dot = 0
        for u, v in zip(cols[a], cols[b]):
            dot += u * v  # step: integer count
        print("columns %s.%s dot product = %d" % ("ABC"[a], "ABC"[b], dot))
alias = []
for a, b, o in ((0, 1, 2), (0, 2, 1), (1, 2, 0)):
    tot = 0
    for i in range(12):
        tot += cols[a][i] * cols[b][i] * cols[o][i]  # step: integer
    alias.append("%s%s on %s %+.3f" % ("ABC"[a], "ABC"[b], "ABC"[o],
                                        tot / 12.0))
print("interaction alias: " + ", ".join(alias))
print("variance of a main effect = 4/12 sigma^2 = 0.333 sigma^2")
print("one factor at a time, 3 baseline + 3 per factor:")
print("  variance = 1/3 + 1/3 = 0.667 sigma^2, twice the matrix")

Single source and citation

Fan, Y. (2026). Cold-Start Dynamics of Short-Video Traffic Pools: Step Thresholds in Completion, Engagement and Replay and a Zero-Ad-Spend Breakout Test Matrix (Version 1.0.0), Chapter 6: The Zero-Ad-Spend Test Matrix: a Twelve-Run Plackett–Burman Design and Its Sample Size. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 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}
}

Other chapter answers in this book