Profit Warning Indicators You Must Track When Operating a Tobacco Content Account

Bottom Line First (Hard Rules From Reviewing My Own Account)

1. 1. Views, likes, and follower growth are not profit indicators; the only things that can sound the alarm are: segmented conversion rate, average order value, repurchase rate, CAC, LTV/CAC, gross margin, and refund rate.
2. 2. When LTV/CAC drops below 1, stop adding ad spend — you are mass-producing losing customers; I use ≥3 as the healthy range, and 1–2 means a yellow light.
3. 3. Tobacco/smoking-cessation content has compliance red lines; if a "good conversion rate" relies on borderline promises, the profit warning must be escalated directly into a survival warning.
4. 4. If average order value spikes short-term while repurchase rate collapses, you are most likely burning trust; a drop in repurchase is more dangerous than a drop in order value.
5. 5. Warnings are not about checking the summary table at month-end, but watching the funnel weekly and unit economics monthly — otherwise you will keep losing money while feeling that "things seem fine."

1. Q3 2024: The Weekly Meeting Where "All Indicators Were Green but Profit Was Red"

In mid-September 2024, in a small meeting room in a shared office in Nanshan District, Shenzhen. I exported the ad backend, the WeCom lead table, and the store orders into three spreadsheets and projected them on the wall.

The surface data looked respectable:

My assistant said at the time: "Conversion looks fine, we got sales."

I recalculated everything by contribution margin:

• Of the 18 orders, 11 were RMB 9.9 info packages, and the margin was almost eaten up by payment channels and customer-service time
• 7 were consultation/solution orders, averaging about RMB 398 collected, but 2 of them were refunded within 7 days
• Rough customer-acquisition cost: 8600 ÷ 1240 ≈ RMB 6.9 per follower, which looked cheap
• Calculated by paying customers, CAC: 8600 ÷ 18 ≈ RMB 478 per paying user
• Of those 18 people, only 2 bought again within the next 90 days

Rough LTV estimate: weighted order value about RMB 160 × 1.1 purchases ≈ RMB 176, against a CAC of 478, giving LTV/CAC ≈ 0.37.

Views were rising, followers were rising, unit economics had already hit the floor.

From that week on, I pinned the "profit warning indicators" on the first page of the weekly report. The framework below is the practical version I polished again in January 2025 — for accounts doing tobacco-harm science, smoking-cessation-pathway content, or oral/respiratory-related science, and converting into the private domain.

2. Use One Table to First Build the Warning System

Common statements in public e-commerce methodology: LTV should cover CAC, LTV/CAC around 3 is comfortable, and below 1 means you are losing money to acquire customers.

On the private-domain side, many health categories emphasize that repurchase should be significantly higher than one-off public-domain transactions — which means: if you only have a first order and no repurchase design, a tobacco-science account can easily become a "high-CAC emotional trash bin."

Numbers can differ by account size; the ratio of red to yellow lights matters more than absolute values. The key is to use the same denominator every week — you are not allowed to use views one week and completion rate the next to swap concepts.

3. Conversion Rate: Break It Down First, Then Sound the Alarm

3.1 Why the "Overall Conversion Rate" Can Deceive You

Many people only look at: transactions ÷ views.

Tobacco content accounts have lots of mixed traffic in the views: people watching cessation stories out of curiosity, people cursing tobacco, minors who mis-tapped, competitors snooping. When the denominator is dirty, conversion always "looks low," so you start changing titles to chase emotional traffic — and it gets dirtier.

I force a three-way breakdown:

1. 1. Content→intent (comment keywords, saves, completion to CTA)
2. 2. Intent→private domain (adding friends/joining groups/leaving contact info)
3. 3. Private domain→paid (info package/course/consultation)

In November 2024, we had a short video "What to do when you can't sleep on day 3 of quitting smoking" with 420,000 views, but only 180 people entered the private domain. My assistant thought conversion was poor.

After breaking it down, we found: comments related to "want to quit" were only a small part of effective interaction — most were emotional resonance like "me too" and "all it takes is willpower."

The problem was not the QR code at the end, but that the topic brought in resonance audiences without filtering out "people ready to act."

The fix: reshoot the same topic with a 15-second self-screening of 3 questions mid-video (daily count, whether you tried quitting in the past 30 days, whether you're willing to keep records), and only give the info package to those who answer "yes" at the end.

The private-domain number dropped from 180 to 95, and paid conversion rose from 2.2% to 7.8%.

The first profit warning: fewer inbound leads but cleaner payers is often a good thing.

3.2 How I Set the Yellow Light for Conversion

Using our January 2025 private-domain funnel as an example (just a reference baseline, not an industry standard):

When conversion drops, I oppose making discounts the first reaction.

Tobacco health users are price-sensitive but even more afraid of being scammed; discounting often brings impulse orders + high refunds, punching through both the order value and refund rate from the next section.

3.3 Response Speed Is a Hidden Conversion Indicator

Around Chinese New Year in February 2025, leads piled up and we only replied the next day.

With the same material, the deal rate for first response within 24 hours was about 2 times higher than over 24 hours (our sample is small, but the direction is stable).

Warning rule: when unanswered <4h messages exceed 20% of total, log a yellow light directly — it disguises itself as "conversion dropping."

4. Average Order Value: Going Up Isn't Always Winning, Going Down Always Demands a Why

4.1 The Typical Form of AOV Being Held Hostage by Lead-Gen Products

We have long maintained three tiers:

In Q4 2024, to chase GMV, operations pushed L1's share to 70%+ of buyers.

Weekly GMV looked good, contribution margin barely moved, and customer-service hours exploded.

The average order value fell from about 260 RMB to 95 RMB — this is the classic path from yellow light to red.

My stance is clear:

• L1 is allowed to exist, but not allowed to be the majority of the deal structure for more than 4 weeks
• The order-value warning looks at structure, not single-week spikes: a continuous decline in the L2+L3 share of buyers matters more than the AOV number

4.2 The Wrong Way to Raise AOV: Forcing High-Price Offers Kills Repurchase

In January 2025, someone suggested pushing 1280 directly to every inbound lead.

We tried it for 9 days: 4 deals, 2 complaints, and the group started labeling us \"you guys are just course sellers.\"

AOV briefly hit 800+, and 90-day repurchase went straight to near zero.

The right path to raise AOV (what we later locked in):

1. 1. L1 delivery must give an executable artifact within 48 hours (not a motivational PDF)
2. 2. On days 3–5, follow up with the user's own check-in data, not mass-sent coupons
3. 3. L2 sells stage problems (sleep, oral-care habits, tracking system, relapse prevention plan), not \"guaranteed quitting\"

After AOV recovered to the 210–280 range, the refund rate finally fell back to an acceptable level.

4.3 AOV Yellow/Red Lights

• Yellow: AOV below the historical median by 20% for 3 consecutive weeks, or L1 share >60%
• Red: deals only via discounts/haggling, with gross margin declining at the same time
• Action: first freeze promotional activities, then check \"is there a mid-tier product to sell,\" rather than opening an even lower price

5. Repurchase Rate: The Life-and-Death Indicator for Tobacco Content Accounts

Public-domain content tends to chase "one viral wave."

People who want profits must accept a fact: cessation/behavior-change users make long decisions with many repetitions — the first order is often just the ticket.

When the industry discusses the private domain, it often mentions that health-category repurchase can be significantly higher than one-off public-domain deals. For our account, I care more about two numbers:

• 30-day repurchase: whether L1 buyers move up to L2
• 90-day repurchase: whether they renew the accompaniment program, or repurchase the toolkit/retrospective service

5.1 Three Real Causes of a Repurchase Collapse (We've Stepped on All of Them)

In October 2024, 90-day repurchase once fell to single-digit percentages.

Reviewing, we found: the content was all "harm scare," users vented emotionally after buying and left — there was no behavioral tool to carry them forward.

Only after shifting 40% of content capacity to "executable checklists + relapse retrospectives" did repurchase slowly return to our acceptable 15%–25% (90 days, including upgrades).

5.2 My Stance

When repurchase is below your own baseline, fix delivery first — don't prioritize more ad spend. Ad spend only amplifies a funnel that never turns back, making CAC look like "loss at scale."

Repurchase warnings:

• Yellow: 90-day repurchase down two weeks in a row month-over-month, or repeat-customer GMV share <20%
• Red: repeat customers near zero with refunds climbing at the same time
• Action: interview 20 non-repeat users for 10 minutes (WeCom voice call is enough) — more useful than filming 5 more viral videos

6. CAC and LTV/CAC: What Decides Whether You Still Dare to Spend

6.1 CAC Must Be Recalculated Using "Paying Customers"

Low follower cost is extremely deceptive.

Tobacco topics easily stir emotional interaction, and cheap followers have a higher share of useless ones.

I watch three CACs at the same time:

1. 1. Follower CAC = promotion spend ÷ new friends added
2. 2. Effective-lead CAC = promotion spend ÷ (people tagged "has quitting intent and is an adult")
3. 3. Paying CAC = promotion spend ÷ new paying customers

In December 2024, one feed-ad plan had follower CAC as low as 4–7 RMB, which looked beautiful; paying CAC touched 500+.

After pausing it, we moved budget to search/keyword materials with stronger intent (e.g., "day N of quitting symptom checklist"), followers cost more, but paying CAC actually dropped.

6.2 How to Estimate LTV Without Fooling Yourself

The rough formula I use:

LTV ≈ average contribution margin × purchases per year × expected lifetime (years)

Or more conservatively: actual contribution margin within 90 days × experience coefficient (1.2–1.8)

Note: use margin, not GMV, to calculate LTV — otherwise you'll fall in love with a toxic structure of high revenue and low margin.

Healthy reference values (combining common e-commerce unit-economics discussions + our practice):

6.3 One "Red-Light Spend Pause" in January 2025

After New Year, a test week: spent 12,000, 22 paying users, paying CAC ≈ 545.

90-day LTV estimated conservatively at 280–320 margin, ratio directly <1.

The weekly meeting decided: stop 3 high-burn materials within 72 hours, cut budget to verification level (daily cap), and go all-in on L2 delivery and old-customer reactivation.

Two weeks later there was no "miracle reversal," but cash outflow was under control — the meaning of profit warning is first to stop the bleeding, not to take off immediately.

7. Three Indicators Often Ignored, Yet Capable of Suddenly Killing Profit

7.1 Refund Rate and Complaint Labels

The easiest pitfall for tobacco-health content is outcome promises.

Once the refund rate is high, recalculating CAC makes it uglier (because the denominator of paying customers is still there, but the margin in the numerator is gone).

Rules:

• Weekly refund rate >10%: freeze external ad spend for that SKU
• When customer-service scripts contain words like "guarantee," "eradicate," "medical substitute": change the script on the spot (compliance is also a profit moat)

E-cigarette/tobacco-related communication has strict boundaries domestically (advertising, online sales, protection of minors, etc.).

Account-level profit warnings must include a "compliance-event probability": one throttle or penalty can invalidate your entire CAC model from the past three months.

7.2 Gross Margin Structure

People who only watch GMV will crash on "peripheral hardware / low-priced physical goods" — shipping, loss, and after-sales punch through the margin.

If a content account touches physical products, I set the gross-margin red light stricter than for pure digital products.

7.3 Headcount Efficiency and Lead Inventory

Leads are not assets — they are ice cubes that expire.

An excessive share of leads not followed up for >7 days is the same as throwing already-spent CAC into the trash.

I require the weekly report to show: lead age distribution. If >7-day age exceeds 30%, log a yellow light.

8. Embed the Warning Indicators Into the Weekly Rhythm (You Can Copy This Directly)

Every Monday, 30 Minutes (Fixed Agenda)

1. 1. Export: spend, new followers, effective intent, deals, refunds, repeat-customer deals
2. 2. Fill the table: four rates (conversion by segment, AOV, repurchase, CAC) + rough LTV/CAC estimate
3. 3. Only yellow/red indicators may be discussed; "viral retrospective" is banned first
4. 4. Output: what to stop this week, what to test, which script to fix (no more than 3 items)

Monthly Unit-Economics Meeting

• Re-estimate LTV (calibrate with real 90-day data)
• Channel ranking: by paying CAC and 30-day margin recovery, not by views
• Product structure: L1/L2/L3 headcount and margin contribution
• Compliance spot-checks: titles, scripts, pinned comments

My "Three Reds Means Stop" Alarm

If any three red lights appear in the same week (e.g., paying CAC red + refund red + repurchase red), automatically trigger a drop of ad budget to verification level — no meeting needed to argue "let's try one more week."

Pre-setting the rule is meant to fight the operator's sunk-cost mentality.

9. Business Judgment Behind the Indicators (Personal View)

1. 1. The biggest profit enemy of a tobacco content account is not \"not knowing how to write titles,\" but feeding wrong confidence with vanity metrics.
2. 2. Conversion should be clean, not high. High conversion on dirty traffic usually relies on wrong promises, and later refunds and word-of-mouth collect everything back with interest.
3. 3. AOV is a structure problem, repurchase is a trust-and-delivery problem, CAC is a channel problem — don't use one prescription (discounting) to cure all diseases.
4. 4. Adding ad spend when LTV/CAC <1 is like looking at losses through a magnifying glass.
5. 5. Compliance is not \"legal's job\" — it is a contingent liability on the profit statement; conversion bought by crossing lines, I would mark directly as a black warning, not yellow.

10. Your 7-Day Execution Checklist

Day 1

Fix the definitions: list your denominators (followers/effective intent/paying), put them in a table, and the team only recognizes this one set.

Day 2

Calculate the three CACs clearly: follower, effective intent, paying; see which one is deceiving you.

Day 3

Break down the last 14 days of deal structure: L1/L2/L3 headcount and margin; if L1 >60%, mark the AOV yellow light.

Day 4

Interview 20 first-purchase users who didn't repurchase, ask three questions: what did they expect, where did it break, what would they still pay for.

Day 5

Set refund-rate and response-speed thresholds; exceed them and automatically throttle ad materials.

Day 6

Estimate an LTV version with 90-day data (must use margin); calculate LTV/CAC and put it at the top of the weekly report.

Day 7

Run a 30-minute warning meeting: only output \"stop/repair/test\" three columns, at most two items each, execute on Monday.

To run a tobacco content account, you can keep chasing views, but please put profit warning indicators above the view count.

Views tell you the volume; conversion rate, AOV, repurchase rate, CAC and LTV/CAC tell you whether you'll still be alive next month.

I would rather have an account with \"fewer inbound leads, healthy unit economics, compounding repurchase\" than one that survives on screenshot highlights at review meetings.