Keeping Profit, Not Just Traffic, for a Tobacco Health Content Account During an Economic Downturn

Key conclusions up front:

1. During a downturn, the most expensive thing isn't traffic — it's the ineffective actions that "still look like they're growing"; plays, likes, and even new followers can all decouple from gross profit.
2. Cutting ad spend isn't about spending less — it's about setting hard rules: "cut if it can't pay back in 7 days, stop if it can't be accounted for in 14 days."
3. In the tobacco health niche, topics that actually close deals mostly fall into "concrete pain + executable next step," not macro scare tactics or pure science lists.
4. If your average order value won't rise, chances are you're selling "single-point reassurance" instead of "staged plans."
5. From late 2024 to 2025, the industry is generally more ROI-conscious and more willing to trim brand campaigns; if you still use the 2021 playbook of "piling on exposure to build the brand" to defend profit, it will only get thinner.

1. The Time We Almost Got Fooled by "Good-Looking" Data

In mid-September 2024, in a small shared office in Nanshan, Shenzhen. On the weekly meeting table lay three screens: the Douyin creator dashboard, the Juliang local-push / feed-ad reports, and the deal records in the WeCom sidebar.

That week the account data "looked great":

• 7-day total plays: about 1.86 million
• New followers: about 4,200
• Private-message consultations: on the order of 310
• Ad spend: about 18,600

In 2022 or 2023, I would have popped the champagne first. But when the finance colleague laid out the gross-profit table, I went quiet: that week contributed about 4,100 in gross profit — and if you count labor, samples, returns, and private-domain customer service, that week was actually performing at a loss, for show.

Where was the problem? We were still using "traffic health" as the KPI. Plays going up means the algorithm hasn't killed you; profit going down means you're feeding a bunch of onlookers who will never buy with attention bought at a high price.

My personal view is blunt: when the economy tightens, an account is not a media outlet — it's a business unit. Media can chase buzz; a business unit must chase "for every 1 yuan spent, can we still keep more than 1 yuan of distributable profit." The "three lows" environment that agencies like Miaozhen keep stressing in 2025 marketing trend reports — growth, demographic dividends, and consumer confidence all weak — translates, for a small team like ours, into: the same 10,000-yuan ad budget that used to buy a wave of consultations now often buys more people who "click and scroll away." Advertisers overall also prefer media with higher ROI and are more willing to cut products for efficiency; if you still fantasize about long-term gains from casting a wide net, reality will grind you into the ground.

So this article is about only one thing: how to defend profit during a downturn instead of defending a pretty data screenshot.

2. Cut Inefficient Ad Spend: Set "Death Clauses" First, Then Talk Optimization

1. Three Kinds of "Fake Optimization" I'm Against

First, watching CPM/CPC every day while ignoring lead cost and deal cost. Cheap CPM may just mean you bought irrelevant audiences; low CPC may just mean clickbait titles that people click and instantly leave.

Second, "let's run it three more days" infinite renewal. In October 2024, we had an oral leukoplakia scare-themed asset. The lead cost looked fine for the first 48 hours, but on days 3–5 the consultation quality fell off a cliff: the WeChat acceptance rate dropped from about 38% to 17%, and the 7-day deal rate after acceptance dropped from about 9% to 2%. Still "observing" is just letting the account bleed.

Third, using total GMV to mask losses in individual campaigns. A winning organic stream holds up the overall number while several ad campaigns keep producing negative gross profit underneath — that's the most common self-deception.

2. The "Three Axes of Stop-Spend" We Actually Use (In Effect Since 2024 Q4)

Pull every campaign / creative unit into the same set of metrics (names can be adjusted to your dashboard; the logic shouldn't change):

Note: warning lines aren't industry scripture; they're what we worked out under a model with an average order value around 199–599 and private-domain sales as the core. If your AOV is 39 yuan or 999 yuan, the numbers must be recalculated; rules can be copied, thresholds can't be blindly copied.

In the second full week of January 2025, we did a "bloody cleanup" under these rules:

• 23 campaigns were running; we shut down 11 (about 48%)
• Daily spend dropped from about 3,200 to 1,900
• Valid consultation volume fell only about 12%
• That week's ad-side gross profit turned from negative to positive, back to roughly a +0.9–1.1 7-day payback ratio range

Some might ask: won't shutting down half hurt algorithm learning? My experience says — it hurts garbage learning. Letting the system keep "learning" from bad creative is teaching it how to waste money more efficiently.

3. The Meeting Where We Called "Stop-Spend": How It Went

Time: the afternoon of January 8, 2025, still in that Nanshan room — one person on ad spend, one on content, one on private domain, and me.

The process had only four steps, 90 minutes total:

1. Export: all campaigns from the past 14 days, fixed fields: spend, impressions, clicks, valid consultations, WeChat adds, deal count, deal amount, refunds.
2. Color-code: 7-day payback < 0.8 → red; WeChat-add cost over the line → yellow; "high plays, low consultations" → its own column.
3. Attribution cross-examination: private-domain staff flipped through chat records on the spot — were the people the red campaigns brought genuinely trying to quit smoking / solve oral problems, or were they "onlookers" drawn in by exaggerated titles?
4. Verdict: red → shut down; yellow → revise creative or targeting, give 48 more hours; green → add budget, but lock the daily budget cap per campaign first so a good campaign doesn't decay after being boosted all at once.

A pitfall we stepped in: early on, we only looked at the "conversion" field in the dashboard and found that a large share of "conversions" were "component clicks" or invalid forms. Since November 2024, valid consultations must be tagged manually or by rules on the private-domain side; dashboard conversions are only a reference. Otherwise you'll optimize toward an audience "great at filling forms but never buying."

4. After Cutting Spend, Cut the "Spend Mindset" Too

I increasingly believe one line: only structures validated by organic traffic deserve to be scaled with paid spend. When budgets tighten in a downturn, you're even less entitled to use paid spend as a lab. The lab should be a small-budget test account or 48 hours of organic data; formal budgets should only scale structures whose "valid lead cost" already meets the bar.

At the industry level, in 2025 advertisers are more competitive on ROI and lean toward performance campaigns — this doesn't mean spend harder, it means admit earlier that some traffic simply can't be bought. Stopping when you can't afford it is more professional than fighting through it.

3. Focus on High-Converting Topics: What Content in a Tobacco Health Account Actually Turns Into Money

In our niche, the line between traffic content and profit content is very clear. I'm giving my judgment from reviewing about 200+ finished videos from August 2024 to February 2025 — not the truth, but a bias verified by deals.

1. High Plays, Low Conversion: Directions I Gradually Do Less

• Pure scare lists: "The 10 biggest harms of smoking" — completion might be fine, but consultations often ask "what about e-cigarettes?" and then leave.
• Macro-policy / industry-news reposts: can grow followers, hard to accumulate demand.
• Cell-level accumulation unrelated to users' lives: without a bridge of "what you'll see in the mirror," professionalism becomes a burden.
• Taking sides, e-cigarette wars: big emotional traffic, big trust damage; serious plans are hard to close in private domain.

2. High Conversion: Topics Closer to "In Pain Now, Willing to Pay to Reduce the Pain"

In our private-domain deal structure, topics that convert better from consultation to payment roughly cluster in these groups:

1. Visible oral damage: gum recession, bad breath, tooth discoloration, abnormal oral mucosa — self-observation (emphasize "go to the hospital"; content only does education and motivation).
2. Quit-failure post-mortems: relapsed after 3 or 7 days, stuck in what scenarios (drinking sessions, overtime, after meals).
3. Concrete management of withdrawal symptoms: irritability, insomnia, cravings — give rhythms and replacement actions, not chicken soup.
4. Family perspective: conversation scripts for spouses and kids persuading a smoker to quit (paying intent is sometimes steadier than the smoker's own).
5. A comparison framework for tools and methods: patches, lozenges, behavioral replacement, nose-inhale aids, etc. — state applicable boundaries and compliance notes clearly, don't package them as miracle cures.

Numeric impression (rough account-side math, not third-party audited): in 2024 Q4, videos like "why the after-meal cigarette is hardest to quit + 7-day replacement actions" might only get 40%–60% of the plays of scare content, but valid lead cost is often 25%–40% lower, and the 7-day deal rate after WeChat add can reach 1.5–2× that of scare content. In a downturn, I pick the latter. Plays are vanity; consultation quality is inventory.

3. The "Skeleton" of a High-Converting Piece of Content (Our Internal Template)

Using "Is your gum line rising because of smoking?" from December 2024 as an example, the structure is:

1. 3-second hook: a scene users relate to (bleeding when brushing, widening gaps between teeth).
2. Mechanism in plain language: touch on the relationship between nicotine and the oral environment, don't pile on jargon.
3. Self-observation steps: what to look at, when to see a doctor — establish a professional boundary.
4. Action fork:

- Already has obvious symptoms → guide to offline medical care

- Wants to reduce tobacco dependence at the same time → guide to private domain for a "quit-smoking behavior log / stage suggestions"

1. Trust patch: state that the content doesn't replace diagnosis/treatment, and products/methods vary by person.

For conversion actions, I advocate singularity. A video that pushes follow, store, livestream, and DM at once is no action at all. We tested: with dual CTAs, the DM rate often drops a notch; with a single CTA "comment 1 to get the log," valid leads are cleaner.

4. The Veto at Topic Selection Meetings

Since February 2025, the topic meeting has one more rule: any topic that can't state "what the viewer will do next after watching" gets vetoed directly. "Gaining knowledge" isn't a next step; "tonight, start counting how long after meals you want to smoke, and record it" is a next step.

4. Raise the Average Order Value: Stop Proving Conversion With Just 9.9 and 39.9

1. The Low-AOV Trap

In the first half of 2024, we were once obsessed with low-price lead generation: 9.9 data packs, 39.9 trial kits. The on-paper conversion rate looked good, then two problems exploded:

• Customer service drowned in low-intent orders, and high-intent users got slower responses;
• The repeat-purchase and upsell path wasn't designed, which is buying a "bargain-hunter user list" with profit.

From March to May, we had a comparison: in months when low-price orders reached 60%+, the per-person service volume of customer service rose about 35%, but per-person gross-profit contribution barely moved. My conclusion: conversion rate is a close relative of vanity metrics; per-person gross profit is the adult metric.

2. How We Built a "Ladder" Instead of Hiking Prices

Average order value isn't raised by shouting a price; it's raised by solution granularity.

A rough ladder (amounts are illustrative; adjust to your compliance and product line):

Three things we insist on when raising the AOV:

1. Diagnose before quoting: smoking years, daily cigarettes, oral discomfort or not, previous quit failures or not — quoting without diagnosis is like a street vendor.
2. What rises is service density, not illusion: the extra money must map to follow-up counts, materials, and exit mechanisms.
3. Let users buy L1: don't squeeze every consultation toward L3. Pushy closing ruins the account's comment section and costs more in the long run.

In November 2024, we tried "directly pushing the 599 plan to everyone." That week the AOV went up, but deal count dropped by nearly half and customer-service conflict scripts multiplied. In December we switched back to "let L1 build a sense of success first, then propose the staged plan on days 5–7," and the monthly average AOV went from about 160 to around 280 (including combos), without refund rates worsening in step. This is the path I think is replicable: trade time for trust, trigger upsells at milestones, instead of leaning on users with scripts.

3. Real Problems in Scripts and Rhythm

Problem one: users immediately ask "do you guarantee success?" Our unified stance: we don't promise quitting results, only the content delivered, follow-up frequency, and tool instructions. Medical promises are a landmine; touch them and you can't hold profit either.

Problem two: price comparisons. Don't compare specs with cheap goods; compare "the time cost of failing once." We get users to calculate: whether three more months of repeated quit-and-relapse cycles (mid) cost more in social and physical wear than a staged plan. People who can make that calculation are easier to close at a higher ticket.

Problem three: family members buying on behalf. A portion of high-ticket deals are paid by spouses. On the content side, doing more topics like "how to talk to your smoking family member" makes the private-domain deal structure much healthier — this is a traffic entry we only really valued in January 2025.

5. Twist Three Things Into One Profit Chain

Cutting spend, picking topics, raising the AOV — if three people each do their own thing, they'll end up blaming each other. We align with one "weekly profit chain":

A Fixed 60 Minutes Every Week (Template)

1. Ad spend: shutdown list + budget reallocation (only add to structures whose valid lead cost meets the bar)
2. Content: this week's 5 topics, each stating "target action / estimated lead-cost band / whether it can be scaled with paid spend"
3. Private domain: this week's main-push tier (L1 or L2), whether upsell-node scripts are updated
4. Review two weeks back: which content brought the highest 14-day gross profit per user — double down on the same structure the week after next

The Shortest Loop From Content Production to Making Money

Organic release → check valid consultations and WeChat-add quality within 48 hours → if the bar is met, scale with a small budget → close in private domain along the diagnostic path → review gross profit at day 14 → decide whether to keep spending long-term or adjust the price structure.

If any link breaks, you're being an influencer, not making profit.

In public trends, marketing budgets are more cautious and more ROI-focused, which is actually good news for small and medium accounts: the era of big players burning money is over; what's left is competing on the precision of unit profit. Tobacco health is also a highly sensitive track, where compliance and trust costs are naturally higher — which means you should spend on "precise" rather than "broad."

6. Seven Days of Actions You Can Start Tomorrow

Day 1

Export campaigns from the past 14 days; flag in red the ones with 7-day payback < 0.8; cut budgets first, don't rush to change the whole account.

Day 2

Define "valid consultation": must include a need tag (quitting motivation / oral issues / family consulting for the smoker, etc.) + a reachable contact method. Without a definition, everything after is empty talk.

Day 3

Classify live content by "scare / list / scenario pain point / method steps," and tally each category's WeChat adds and deals over the past 30 days. Your own data will slap your face — that's a good thing.

Day 4

Stop updating 2 series that "only grow plays, not consultations"; use the saved filming time to polish 1 scenario-pain-point video with a single CTA.

Day 5

Draw the L0–L2 product ladder, clarify each tier's deliverables; delete high-price promises that "can't state the delivery."

Day 6

Sample 20 unconverted chats in private domain and tag the sticking points: price, trust, unclear needs, just small talk. If price is the sticking point, optimize the ladder; if trust, add content evidence and compliant phrasing.

Day 7

Hold the 60-minute profit-chain weekly meeting and make only one decision: which 3 campaigns / which 5 pieces of content get the money from Monday to Sunday next week. Fewer decisions than that means you're still spreading effort evenly.

Conclusion: The Real Moat in a Downturn

I've seen too many accounts in the tight years of 2024–2025 shouting "traffic is expensive" while unwilling to stop familiar campaigns, unwilling to give up frightening viral structures, unwilling to step out of the 39.9 comfort zone. What's expensive isn't traffic — it's path dependence.

If a tobacco health content account still wants to make money, it has to accept a slightly cruel switch:

• From "I want more people to see" to "I want more of the right people to see";
• From "I want more likes" to "I want more people to complete the next step";
• From "I want more orders" to "I want the gross profit and replicability behind every order."

Defending traffic is defending yesterday's habits; defending profit is defending your right to survive the next quarter. If your back-end plays are still rising and your wallet is already thinning, don't add budget first — take the knife and take a cut at low-efficiency spend, low-converting topics, and low-margin product shelves. After the cut, you'll often see for the first time: what truly makes the account money has always been just a few pieces of content, a few campaigns, and a slice of the AOV.