What does The Nightly Funnel conclude, in numbers, about “The Funnel Check-up: Six-Step Beta Posteriors, Nightly Change Priorities and a Compliant Candidate Pool”?
Answer: Night B uses one night's small counts: 1,650 people enter, followed by 1,010, 445, 190, 112 and 77. Now the sample step has P(top) of only 0.714 and the first screen 0.286. The sample step's priority interval runs from 3,511 to 5,140 and the first screen's from 3,270 to 4,968, with medians of 4,305 and 4,067; they really cannot be told apart. The rule returns HOLD: change nothing tonight, add another night of data and recompute tomorrow.
Scope of the numbers: Its parameters are illustrative; all calculations are generated by its 23 programs; it is general business analysis, not business, legal or financial advice, and it promises no rise in conversion.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 0.286, 0.714, 1010, 112, 1650, 190
| night A week pooled: 12,000, 7,440, 3,350, 1,410, 820, 560 | |||
| step | P(top) | priority | interval 2.5% to 97.5% |
| visit | 0.000 | 14,401 | 13,240 to 15,641 |
| hold | 0.000 | 28,678 | 26,458 to 30,967 |
| sample | 1.000 | 32,323 | 30,055 to 34,646 |
| tier | 0.000 | 16,907 | 15,625 to 18,249 |
| checkout | 0.000 | 10,931 | 9,842 to 12,043 |
| decision: CHANGE sample (P(top) 1.000 vs bar 0.80) | |||
| night B one night: 1,650, 1,010, 445, 190, 112, 77 | |||
| step | P(top) | priority | interval 2.5% to 97.5% |
| visit | 0.000 | 2,030 | 1,614 to 2,520 |
| hold | 0.286 | 4,067 | 3,270 to 4,968 |
| sample | 0.714 | 4,305 | 3,511 to 5,140 |
| tier | 0.000 | 2,242 | 1,788 to 2,731 |
| checkout | 0.000 | 1,480 | 1,103 to 1,900 |
| decision: HOLD (P(top) 0.714 vs bar 0.80) |
Key formula
Show the full program
# Chapter 1. p12: nightly change priority = lost visitors x downstream
# expected net contribution, with posterior uncertainty.
# priority_k = n_k * (1 - p_k) * NET * prod_{j > k} p_j, where p_k is a
# draw from Beta(1 + pass, 1 + drop). NET (net contribution per paid
# order, in illustrative units) is fixed. The hold rule: change a step
# tonight only if P(step has the largest priority) >= 0.80; otherwise
# hold, because the priority intervals overlap too much.
import random
STEPS = ["visit", "hold", "sample", "tier", "checkout"]
NET = 42.0
NIGHTS = {"A week pooled": [12000, 7440, 3350, 1410, 820, 560],
"B one night": [1650, 1010, 445, 190, 112, 77]}
DRAWS = 20000
PASS_P = 0.80
rng = random.Random(40512)
def q(xs, f):
xs = sorted(xs)
return xs[int(f * (len(xs) - 1))]
for name in sorted(NIGHTS):
c = NIGHTS[name]
wins = [0] * len(STEPS)
pri = [[] for _ in STEPS]
for _ in range(DRAWS):
p = [rng.betavariate(1 + c[k + 1], 1 + c[k] - c[k + 1])
for k in range(len(STEPS))]
row = []
for k in range(len(STEPS)):
down = NET
for j in range(k + 1, len(STEPS)):
down *= p[j]
row.append(c[k] * (1 - p[k]) * down)
best = max(range(len(row)), key=lambda k: row[k])
wins[best] += 1
for k in range(len(STEPS)):
pri[k].append(row[k])
print("night {}: {}".format(name, ", ".join(
"{:,}".format(x) for x in c)))
print(" step P(top) priority interval 2.5% to 97.5%")
for k in range(len(STEPS)):
print(" {:<9} {:.3f} {:>9,.0f} {:,.0f} to {:,.0f}".format(
STEPS[k], wins[k] / DRAWS, q(pri[k], 0.5),
q(pri[k], 0.025), q(pri[k], 0.975)))
top = max(range(len(STEPS)), key=lambda k: wins[k])
ptop = wins[top] / DRAWS
verdict = "CHANGE " + STEPS[top] if ptop >= PASS_P else "HOLD"
print(" decision: {} (P(top) {:.3f} vs bar {:.2f})".format(
verdict, ptop, PASS_P))Single source and citation
Fan, Y. (2026). Autonomous Nightly Funnel Micro-Surgery, Bandit Traffic Allocation and Zero-Touch Conversion Self-Improvement (Version 1.0.0), Chapter 1: The Funnel Check-up: Six-Step Beta Posteriors, Nightly Change Priorities and a Compliant Candidate Pool. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-405 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026405,
author = {Fan, Yafu},
title = {Autonomous Nightly Funnel Micro-Surgery, Bandit Traffic Allocation and Zero-Touch Conversion Self-Improvement},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 2 · Thompson Sampling and Regret: from Two Small Samples in 1933 to the Even Split
- Chapter 3 · The Reward Function: Net Contribution per Visitor, Not the Conversion Rate
- Chapter 4 · Non-stationary Traffic: Weekday Effects, Discounted Memory and Sliding Windows
- Chapter 5 · Elimination Rules: Minimum Exposure, Posterior Win Probability and Time-uniform Confidence Sequences
- Chapter 6 · Trustworthy Data: Negative Bias in Adaptive Data, Sample Ratio Mismatch and Guardrail Rollback
- Chapter 7 · Classical Sources East and West: "At the Limit, Change" in the Great Commentary, the Sunzi on Weak and Strong Points, and Random Allocation in Peirce and Jastrow
- Chapter 8 · The Nightly Funnel Ledger: Field Tools and Checker Specifications, a Pre-registered Test and the Conditions of Refutation