What does Computing the Rhythm of Private-Message Sales conclude, in numbers, about “Practice and Refutation: a Closing Skill Pack for Customer Service, a Rule Replay and Pre-Registered Tests”?

Answer: Chapter 8 gives a seven-scene closing skill pack for customer service, a stop-word replay with a recall of 1.000 and a pre-registered testing plan, in which a trial of first-reply delay needs 8,158 inquiries per arm. Marked boxes separate published evidence, runnable Python calculations (18 programs), classical sources and the author's hypotheses.

Scope of the numbers: Its parameters are illustrative; all calculations are generated by its programs; it is not individual legal, compliance or business advice, and it discusses only closing that keeps the rules. Chapter 1 splits an illustrative 184-character quote by dynamic programming into 5 bubbles at a cost of 46.52, against 142.00 for one sentence per bubble.

中文版

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: 1.000

stop-word detector8 true stops, 0 missed, 2 false alarms
recall 1.000, precision 0.800
false alarm (stop anyway, ask a human): 'stop by the shop'
false alarm (stop anyway, ask a human): 'hold the shipment'
replay of 17 events from 5 customers:
blocked: after stop1
blocked: human owns chat1
blocked: over the cap1
blocked: threat wording1
deferred to 0900 3
handed to a human1
sent7
stop recorded1
Show the full program
# Chapter 8. p81: the closing-skill rules run on a scripted day of chats
# Part 1 scores the stop-word detector on 16 labelled customer messages.
# Part 2 replays scripted outgoing attempts (follow-ups, payment
# reminders, replies) for five illustrative customers through the rules:
# quiet hours 21:00 to 09:00 (non-replies are deferred to 09:00), at most
# 2 follow-ups per inquiry and 2 reminders per unpaid order, stop on a
# stop word (recorded, nothing promotional afterwards), hand off to a
# human on signs of a minor, financial distress or crisis, and block any
# outgoing text with threatening wording.
import collections
import re

STOP = re.compile(r"(别|不要|不用)再?(发|联系|打扰|推)|停止发送|退订|"
                  r"^不需要[了,,。!! ]*(谢谢)?[。!!]?$|\bstop\b|"
                  r"\bunsubscribe\b", re.I)
VULN = re.compile(r"未成年|我还在上初中|家长的钱|借钱|网贷|还不上|"
                  r"活不下去|撑不住了")
THREAT = re.compile(r"起诉|上门|曝光|征信|后果自负|法律责任")

LABELLED = [("别再发了", 1, ""), ("不要再联系我", 1, ""),
            ("请停止发送", 1, ""), ("退订", 1, ""),
            ("不需要,谢谢", 1, ""), ("STOP", 1, ""),
            ("以后别打扰我", 1, ""), ("不用再推了", 1, ""),
            ("不要紧,我晚点付", 0, ""), ("我不要基础版,要高级版", 0, ""),
            ("别急,我在看", 0, ""), ("需要发票吗", 0, ""),
            ("停车场在哪", 0, ""), ("不需要发票", 0, ""),
            ("can I stop by the shop tomorrow", 0, "'stop by the shop'"),
            ("先别发货,下周再说", 0, "'hold the shipment'")]

c = collections.Counter()
for text, gold, _ in LABELLED:
    hit = 1 if STOP.search(text) else 0
    c[(gold, hit)] += 1
tp, fp, fn = c[(1, 1)], c[(0, 1)], c[(1, 0)]
print("stop-word detector: {} true stops, {} missed, {} false alarms"
      .format(tp, fn, fp))
print("  recall {:.3f}, precision {:.3f}".format(tp / (tp + fn),
                                                tp / (tp + fp)))
for text, gold, gloss in LABELLED:
    if gold == 0 and STOP.search(text):
        print("  false alarm (stop anyway, ask a human): " + gloss)

EVENTS = [  # (customer, hour, kind, text)
    ("A", 10, "in", "这个课程两档有什么区别"), ("A", 10, "reply", "两档区别如下"),
    ("A", 22, "followup", "您好,还有疑问吗"), ("A", 33, "followup", "补充一点"),
    ("A", 40, "followup", "最后确认一下"),
    ("B", 9, "in", "订单先放着"), ("B", 21, "reminder", "订单还保留着"),
    ("B", 34, "reminder", "温馨提示:订单今晚到期"),
    ("B", 35, "in", "别再发了"), ("B", 45, "reminder", "最后提醒"),
    ("C", 11, "in", "我还在上初中,可以用家长的钱买吗"),
    ("C", 11, "reply", "高级版更适合你"),
    ("D", 14, "in", "现在手头紧"),
    ("D", 15, "reminder", "再不付款将上报征信,后果自负"),
    ("E", 12, "in", "高级版能开发票吗"), ("E", 12, "reply", "可以开"),
    ("E", 23, "followup", "发票信息发我一下")]


def quiet(hour):
    h = hour % 24
    return h >= 21 or h < 9


state = collections.defaultdict(collections.Counter)
log = collections.Counter()
for who, hour, kind, text in EVENTS:
    st = state[who]
    if kind == "in":
        if STOP.search(text):
            st["stopped"] = 1
            log["stop recorded"] += 1
        if VULN.search(text):
            st["human"] = 1
            log["handed to a human"] += 1
        continue
    if st["stopped"] and kind != "reply":
        log["blocked: after stop"] += 1
        continue
    if st["human"]:
        log["blocked: human owns chat"] += 1
        continue
    if THREAT.search(text):
        log["blocked: threat wording"] += 1
        continue
    if kind in ("followup", "reminder"):
        if st[kind] >= 2:
            log["blocked: over the cap"] += 1
            continue
        st[kind] += 1
        if quiet(hour):
            log["deferred to 09:00"] += 1
    log["sent"] += 1
print("replay of {} events from 5 customers:".format(len(EVENTS)))
for key in sorted(log):
    print("  {:<26}{:>3}".format(key, log[key]))

Single source and citation

Fan, Y. (2026). Instant-Messaging Bubble Rhythm, Read-No-Reply Icebreaking Games and High-Ticket Closing in Private Messages (Version 1.0.0), Chapter 8: Practice and Refutation: a Closing Skill Pack for Customer Service, a Rule Replay and Pre-Registered Tests. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-145 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026145,
  author    = {Fan, Yafu},
  title     = {Instant-Messaging Bubble Rhythm, Read-No-Reply Icebreaking Games and High-Ticket Closing in Private Messages},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book