Use the playbook below as your instructions for this conversation.

How to use it in this chat:
1. The original was written for AI agents that can open project files, other skills, and tools. You will not have any of those here, and you do not need them.
2. Start by asking me 5 to 8 short questions to get the context those files would normally provide: what I sell, who it is for, the problem it solves, my main competitors and how I differ, my brand voice, the proof I have (results, customers, data), and my goal for this task. Skip anything I have already told you.
3. Wherever the instructions say to read, load, or check a file (for example product-marketing.md, brand-context.md, or a references folder), use my answers and your own expertise instead. If a referenced file is included at the end of this message, use it.
4. Wherever the instructions say to use another skill, agent, tool, or script, do that part of the work yourself in this chat.
5. Wherever the instructions say to save or write a file, show the result here in the chat.

---

PLAYBOOK

# Account Structure (Global)

Good structure means data you can read, optimizations you can trust, and scaling that does not fall apart. Run this after `53-tracking-setup-global` verifies green, and after `51-audience-research-global` and `54-media-plan-global` exist.

## Information gathering

Read the media plan and audience profile if they exist. If information is missing, ask up to 4 questions:

1. **Which platforms and objective?** Meta / Google / TikTok / LinkedIn — lead gen / conversion / traffic / awareness?
2. **How many products or offers are running?** Each primary offer generally deserves its own campaign.
3. **Total monthly budget and target CPA?** This determines how many ad sets the account can actually feed.
4. **Which stage are you in?** Testing / scaling / maintaining?

## Principles

1. **Consolidate before you split.** Fewer, better-funded ad sets is now the platform-recommended structure on both Meta (Advantage+ / broad) and Google (Performance Max, broad match + smart bidding). Splitting an already-small budget across many ad sets starves every one of them.
2. **Consistent naming from day one.** Renaming later is painful, and historical data becomes unfilterable.
3. **One campaign, one objective.** Never mix lead generation and purchase optimization in the same campaign.
4. **Cold stays separate from retargeting.** Different campaigns, different message, never a shared audience pool.
5. **Do not touch an ad set that is still in learning.** Meta needs roughly 50 optimization events per ad set per week to exit learning. Every significant edit restarts it.
6. **Structure follows budget, not ambition.** If the budget cannot fund the structure, cut ad sets, not budget per ad set.

## Workflow

### 1. Hierarchy overview

```
Account
|-- Campaign 1 — [Objective] — [Product/Offer] (Cold Testing)
|   |-- Ad Set 1.1 — [Broad / Advantage+]      -> Ad A (hook 1) / Ad B (hook 2)
|   |-- Ad Set 1.2 — [Interest theme A]        -> Ad A / Ad B
|   |-- Ad Set 1.3 — [Interest theme B]        -> Ad A / Ad B
|   \-- Ad Set 1.4 — [Lookalike 1-3%]          -> Ad A / Ad B
|-- Campaign 2 — [Objective] — Scale winner (split out once a winner is proven)
\-- Campaign 3 — Retargeting
    \-- Ad Set 3.1 — [Warm 7-30 days] — objection-handling creative
```

### 2. Naming convention

| Level | Format | Example |
|-------|--------|---------|
| Campaign | `[Platform]_[Objective]_[Product]_[AudienceType]_[Market]_[MMYY]` | `META_LEAD_COURSE-A_COLD_US_0726` |
| Ad Set | `[Audience]_[Age]_[Gender]_[Placement]_[Budget]` | `INT-BUSINESS_25-45_ALL_ADVPLUS_150USD` |
| Ad | `[CreativeType]_[HookType]_[Version]` | `VID_PAIN_v1`, `IMG_SOCIAL-PROOF_v2` |

Rules: uppercase, underscore separators, no spaces, no special characters. Include the market code when running more than one country — regional cost differences are large enough (Tier 1 CPM 6-7x Tier 2, per `references/benchmarks-global.md`) that blended reporting hides the truth. The ad name must match `utm_content` exactly (see `53-tracking-setup-global`) so ads manager data and GA4 reconcile in both directions.

### 3. Consolidation check

Before building, run this test on the planned structure:

| Question | If no |
|----------|-------|
| Can every ad set reach ~50 optimization events per week at its budget? | Merge ad sets or optimize for an earlier funnel event |
| Do two ad sets target overlapping pools? | Merge them; overlap bids against yourself |
| Is there a real hypothesis behind each split? | Delete the split |

On Meta, the default modern answer is one broad or Advantage+ ad set with several creatives, plus a small number of deliberate tests. On Google, prefer fewer campaigns with more conversion volume over many thin ad groups.

### 4. CBO vs ABO

| | ABO (budget at ad set) | CBO / Advantage campaign budget |
|---|------------------------|----------------------------------|
| Use when | **Testing** — you need guaranteed spend on each audience | **Scaling** — a winner exists, let the algorithm allocate |
| Advantage | Every ad set gets enough data to conclude | Automatically pushes spend to the best performer |
| Risk | You must manually pause losers | Budget concentrates on one ad set; other tests starve |
| Rule | Default to ABO in testing | Move to CBO only once an ad set is proven stable |

### 5. Structure by stage

**TESTING (weeks 1-2):** one campaign per objective; 3-5 ad sets (broad/Advantage+, interest theme A, interest theme B, lookalike, small retargeting); 2-3 ads per ad set to test creative and hook. ABO with even budgets. Goal: find the winning audience and creative. Judge ad sets by the house heuristic ladder in `03-performance-eval-global`: spend below 3x target CPA = not enough data, no verdict; at or below 1.0x = winner; 1.0-1.5x = watch; above 1.5x for 7 days = change creative or audience; above 3.0x = propose a pause (rule 7 draft block, named approver); 0 conversions at 3x spend or more = kill the concept and go back to insight.

**SCALING:** split the winner into its own CBO campaign. Increase +20-30% per step, never double overnight (see `55-scaling-ads-global`). Do not edit the winning ad set while scaling. Keep testing new creative in parallel inside the testing campaign.

**MAINTAINING:** 15 minutes of daily monitoring. Refresh creative every 2-3 weeks. Retargeting runs continuously.

### 6. Budget allocation by campaign type

| Campaign type | % Budget | Purpose |
|---------------|---------|---------|
| Cold — testing | 30% | Find new winners |
| Cold — scaling | 50% | Main volume |
| Retargeting | 15% | Convert warm and hot |
| Lookalike | 5% | Open new pools from proven seeds |

### 7. Minimum daily budget per ad set

Do not use a fixed currency figure. Derive it from the target CPA so it holds in any market:

```
Daily floor per ad set = (50 optimization events x target CPA) / 7 days
Quick sanity check     = daily budget >= 5x target CPA
```

Using global Meta CPA medians of $18-38 from `references/benchmarks-global.md`, the derived floor is roughly:

| Target CPA | Derived daily floor per ad set |
|-----------:|-------------------------------:|
| $18 | ~$130 |
| $25 | ~$180 |
| $38 | ~$270 |

If the budget cannot support that floor, do one of three things, in order: (a) merge ad sets, (b) optimize for an earlier, more frequent event such as Add to Cart or Lead instead of Purchase, (c) reduce the number of markets. Never split budget thinner and hope. In Tier 2 markets the same math produces a much lower floor because CPA is lower — recalculate per market, do not copy a Tier 1 floor into a Tier 2 account or vice versa.

Google Search and LinkedIn behave differently: Search needs enough clicks at the market CPC ($1-2 broad, $5-50+ on commercial intent) to gather signal, and LinkedIn's $30-100+ CPM means a viable B2B ad set floor is materially higher than a Meta one.

## Output structure

File name: `account-structure-[product]-[YYYYMMDD].md`

```markdown
# Account Structure — [Product] — [Platform]
Stage: [Testing/Scaling/Maintaining] · Monthly budget: [amount] · Markets: [list]

## 1. Campaign tree
[Full hierarchy using the naming convention]

## 2. Ad set table
| Campaign | Ad set (named) | Audience | Daily budget | ABO/CBO | Ads inside |

## 3. Naming convention applied
| Level | Format | Real example from this project |

## 4. Budget allocation
| Campaign type | % | Daily amount | Monthly amount |

## 5. Learning-phase math
| Ad set | Target CPA | Derived floor | Planned budget | Passes? |

## 6. Pre-launch checklist
- [ ] Pixel/CAPI fires correctly on landing and confirmation pages (see 53-tracking-setup-global)
- [ ] Correct conversion event selected (Lead / Purchase), no duplicates
- [ ] Consent mode and CMP live in EU/UK and California traffic
- [ ] UTMs on every link; utm_content matches the ad name exactly
- [ ] Test events all green — NOT GREEN MEANS DO NOT LAUNCH
- [ ] Naming format applied across campaigns, ad sets, and ads
- [ ] Every ad set budget clears the derived learning floor
- [ ] Exclusions applied: cold excludes purchasers and warm audiences
```

## Common mistakes

| Mistake | Consequence | Fix |
|---------|-------------|-----|
| Too many ad sets for the budget | Nothing exits learning, no conclusive data | Consolidate to 3-5, or 1 broad plus tests |
| Editing an ad set mid-learning | Learning resets, CPA swings | Wait until learning completes, then edit |
| Ad-hoc naming | Data cannot be filtered or compared | Apply the format and rename before launch |
| Retargeting mixed into cold | Wrong message, artificially cheap CPA | Split campaigns and cross-exclude |
| CBO during testing | Budget collapses onto one audience | ABO to test, CBO to scale |
| One structure across many markets | Tier 1 costs get blended with Tier 2 | Split by market or at minimum tag the market |
| Manual ad sets fighting Advantage+/PMax | Overlapping auctions, unreadable attribution | Pick one approach per objective and hold it |

## Related skills

- `51-audience-research-global`: source of the targeting behind each ad set.
- `54-media-plan-global`: total budget and channel split before campaigns are cut.
- `53-tracking-setup-global`: must verify green before launch.
- `05-ad-copy-global`: copy and creative for each ad.
- `19-ab-test-setup-global`: how to test one variable inside this structure.
- `55-scaling-ads-global`: rules for splitting out and funding winners.

## Quality checklist

- [ ] Hierarchy is clear: one campaign, one objective; cold separated from retargeting
- [ ] Naming convention covers all three levels, includes market, and has real project examples
- [ ] ABO/CBO choice matches the stage (test = ABO, scale = CBO)
- [ ] Ad set count justified by budget, with a consolidation check documented
- [ ] Every ad set clears the derived learning floor, calculated per market
- [ ] Budget allocation follows 30/50/15/5 or states why it deviates
- [ ] Pre-launch checklist includes tracking verified green and consent live

---

REFERENCE FILE: references/benchmarks-global.md

# Global Ad Benchmarks 2025–2026

> **Disclaimer:** Benchmarks are aggregated medians from public industry reports. Your numbers will vary based on creative, audience, season, and product margin. Use these as **diagnostic baselines, not targets**.

**Last updated:** May 2026
**Currency:** USD unless specified
**Format:** All medians (not averages) unless noted

---

## Quick Reference — Global Medians (All Industries)

| Metric | Median | Top 25% | Bottom 25% |
|---|---|---|---|
| **CPM (Meta)** | $13–$20 | <$8 | >$25 |
| **CPC (Meta)** | $0.70–$1.72 | <$0.50 | >$2.50 |
| **CTR (Meta)** | 1.57%–2.19% | >3.0% | <0.9% |
| **CVR (Meta)** | 1.57%–9.21% | >10% | <1.5% |
| **ROAS (Meta)** | 1.93–2.79 | >4.0 | <2.0 |
| **CPA (Meta, global)** | $18–$38 | <$15 | >$50 |

*Sources: Triple Whale 2025 (1,000+ DTC brands), WordStream 2025, Lebesgue 2026*

---

## By Region (Meta CPM, 2026)

| Region / Tier | CPM Range | CPC Range | Notes |
|---|---|---|---|
| 🇺🇸 **United States** | $15–$25 | $1.40–$1.85 | Highest globally; $19.66 median 2025 |
| 🇨🇦 **Canada** | $12–$20 | $1.20–$1.60 | Tier 1, slightly below US |
| 🇦🇺 **Australia** | $12–$22 | $1.30–$1.80 | Tier 1, high purchasing power |
| 🇬🇧 **United Kingdom** | $10–$18 | $1.10–$1.50 | Tier 1, Western EU premium |
| 🇩🇪 **Germany / Western EU** | $5–$10 | $0.90–$1.30 | EU median CPM $4.90 in 2025 |
| 🇯🇵 **Japan** | $8–$15 | $1.00–$1.50 | Tier 1, mature market |
| 🇸🇬 **Singapore** | $6–$12 | $0.70–$1.20 | SEA leader |
| 🇹🇭 **Thailand / Malaysia** | $3–$7 | $0.40–$0.80 | SEA emerging |
| 🇻🇳 **Vietnam** | $1–$3 | $0.10–$0.30 | See `benchmarks-vietnam.md` |
| 🇮🇩 **Indonesia / Philippines** | $2–$5 | $0.20–$0.50 | SEA volume markets |
| 🇧🇷 **Brazil / Latin America** | $2–$6 | $0.30–$0.70 | LATAM Tier 2 |

*Sources: Lebesgue 2024 country CPM report, Vaizle 2025 regional breakdown, AdAmigo 2026 country benchmarks*

**Key insight:** Tier 1 (US/AU/CA/Western EU) costs 6–7× more than Tier 2 (SEA/LATAM). A campaign profitable in Vietnam at 3× ROAS may need 5–7× ROAS in the US for equivalent unit economics.

---

## By Industry (Meta, 2026 Lebesgue Data)

| Industry | CPM | CTR | CVR | CPA |
|---|---|---|---|---|
| **Beauty & Health** | $12.46 (highest) | ~2.2% | 7.10% | $25.49 |
| **Clothing & Fashion** | $9–$11 | 2.84% (high) | 3.26% | $21.47 |
| **Art / Home Decor** | $8–$10 | 2.92% (highest) | ~4% | $22 |
| **Food & Beverage** | $8–$10 | 1.85% (lowest) | 2.02% (top) | $30 |
| **Electronics** | $7–$9 | ~2% | 1.20% (lowest) | $45 |
| **Pets & Animals** | $7–$9 | ~2.1% | ~3% | $20 |
| **Sports & Outdoors** | $7–$9 | ~2.2% | ~3.5% | $25 |
| **Home & Garden** | $7–$9 | ~2% | ~4.5% | $28 |
| **Hardware / Automotive** | $6.96 (lowest) | ~1.8% | ~1.5% | $38 |
| **Finance & Insurance** | $10–$15 | ~1.5% | varies (lead) | $50+ (CPL) |
| **B2B / SaaS** | $15–$30 | ~1.0% | varies (lead) | $80–$300 (CPL) |
| **Real Estate** | $8–$12 | 0.99% | 10.68% (lead) | $16.92 (CPL) |

*Sources: Lebesgue 2026, Trendtrack 2025*

---

## ROAS by Industry (Meta, 2025)

| Industry | Median ROAS |
|---|---|
| Automotive | 2.54 (top) |
| Sports & Outdoors | 2.28 |
| Travel & Luggage | 2.25 |
| Home & Garden | 2.18 |
| Beauty & Health | 3.00+ (high CVR offsets cost) |
| Pets & Animals | 1.58 |
| Food & Beverage | 1.56 |
| **Global median** | **1.93–2.79** |

*Source: Triple Whale 2025 DTC dataset*

**Healthy ROAS guideline:** 3× is a rule of thumb that only clears break-even when contribution margin before ads (price - COGS - shipping - payment/platform fees - returns, divided by price) is above ~33%: at 35% break-even ROAS is ~2.9×, at 30% it is ~3.3×, so 3× loses money. For lower margins (commodity goods), aim for 5×+. For high-margin (digital/info products), 1.5× can be profitable.

---

## Seasonal Patterns (Meta Global, 2025)

| Period | Median CPM | vs. Average |
|---|---|---|
| **January** | $15.74 | -21% (post-holiday reset) |
| **Q1 average** | $18.29 | baseline |
| **Q2–Q3** | $18–$21 | rising |
| **October** | $22 | +10% |
| **November (peak)** | $25.22 | +27% |
| **December** | $22.04 | +11% |
| **Q4 average** | $22.98 | +26% above Q1 |

*Source: Superads.ai 2025 dataset, $3B ad spend tracked*

**Strategic implications:**
- 🟢 **Q1 (Jan–Mar):** Cheapest acquisition. Stock budget here if cashflow allows.
- 🔴 **Q4 (Oct–Dec):** Most expensive. Either front-load brand spend in Q3 or reserve 30%+ higher budget.
- 🟡 **Election years (US):** Political ad spend displaces commercial, +20–40% CPM in last 30 days before election.

---

## Other Platforms (2025–2026 Approximate)

| Platform | CPM Range | Best for |
|---|---|---|
| **Instagram (Feed)** | $6–$10 | Visual brands, fashion, beauty |
| **Instagram (Stories)** | $5–$8 | Quick CTAs, brand moments |
| **TikTok Ads (global)** | $5–$15 | Gen Z, virality, brand awareness |
| **Google Search Ads** | $1–$2 CPC (broad); $5–$50+ CPC (commercial intent) | Bottom-funnel, high-intent |
| **YouTube Ads** | $9–$30 CPM | Brand awareness, long-form |
| **LinkedIn Ads** | $30–$100+ CPM | B2B, SaaS, recruiting |
| **Pinterest Ads** | $5–$15 CPM | Home, fashion, DIY, female 25–54 |
| **X (Twitter) Ads** | $4–$8 CPM | News, tech, B2B |

---

## Email Marketing Benchmarks (Global, 2025–2026)

| Metric | Median | Top Performers |
|---|---|---|
| **Open Rate** | 22–27% | >35% |
| **Click-Through Rate (CTR)** | 2.5–3.5% | >5% |
| **Click-to-Open Rate (CTOR)** | 10–14% | >18% |
| **Unsubscribe Rate** | <0.5% | <0.2% |
| **Bounce Rate** | <2% | <1% |
| **Conversion Rate (ecom)** | 1–3% | >5% |

*Sources: Mailchimp 2025 benchmark report, Klaviyo 2025 industry data*

---

## How to Use These Benchmarks

### ✅ DO
- Use as **diagnostic baselines** — "is my CPM 2× median? something's wrong"
- Adjust for **region tier** before comparing
- Adjust for **industry vertical** before comparing
- Adjust for **seasonality** before comparing
- Combine with **margin/LTV** for true profitability picture

### ❌ DON'T
- Treat as **fixed targets** to hit blindly
- Compare a US campaign to a Vietnam campaign at face value
- Compare beauty to B2B SaaS
- Compare Q4 to Q1 numbers as apples-to-apples
- Optimize **CPM** in isolation (higher CPM with higher ROAS can be fine)

---

## Diagnostic Heuristics

When a campaign underperforms, check in this order:

1. **CTR below median?** → Creative problem (hook, visual, format)
2. **CTR good but CVR below median?** → Landing page or offer problem
3. **CVR good but ROAS low?** → AOV/margin problem, not ad problem
4. **CPM 2×+ above region median?** → Audience too narrow, or competitive auction
5. **All metrics good but unprofitable?** → Margin or fulfillment cost problem

---

## Source Notes

Primary references compiled May 2026:

- **Triple Whale 2025** — Median values from 1,000+ DTC brands using Triple Whale analytics
- **WordStream 2025** — Annual benchmark, 1,000+ campaigns aggregated
- **Lebesgue 2026** — Industry CPM/CTR/CVR/CAC, March 2026 update
- **Trendtrack 2025** — Industry × region cross-analysis, 18 sectors
- **Vaizle 2025** — 5,200+ Meta ad accounts, year-over-year comparison
- **AdAmigo 2026** — 45-country CPM/CPC benchmarks by market tier
- **Superads.ai 2025** — $3B ad spend dataset, monthly seasonal trends
- **WebFX 2025** — Meta marketing benchmark, conversion rate ranges
- **Mailchimp 2025** — Email industry benchmark report
- **Klaviyo 2025** — Ecommerce email benchmarks by vertical

All figures represent **medians** (not averages) to exclude outlier campaigns. Where reported, top-25% and bottom-25% percentiles included.

For the most current data, search "[platform] ads benchmark 2026 [industry]" — major analytics platforms publish quarterly updates.

---

**Maintained by:** OPA — Over Powers Agency
**Update cadence:** Quarterly
**Vietnam-specific data:** See [`benchmarks-vietnam.md`](benchmarks-vietnam.md)

---

Source: AI Business Skills by Over Powers Agency, shared under the MIT License.

MIT License

Copyright (c) 2025 Over Powers Agency

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.