ワンクリックで
campaign-optimization
Campaign optimization methodology — budget allocation, creative testing, scaling playbook, week-by-week ops
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Campaign optimization methodology — budget allocation, creative testing, scaling playbook, week-by-week ops
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Email marketing automation — drip sequences, deliverability, template patterns, provider APIs (Resend, SendGrid, Postmark)
Funnel diagnostics engine — bottleneck detection, cross-data correlation, AARRR benchmarks by vertical
High-Velocity Advertising (HVA) — the discipline for creative discovery at machine speed. CLEAR decision rule, the five-layer Stack, the Read Ladder, the two-clocks asymmetry, Benjamini-Hochberg correction, the Vault moat, and the Fit Boundary. The spec behind /hva, /hva-forge, /hva-lint, /hva-review, /hva-vault and the hva-desk agent.
Direct response copywriting framework for paid ads — Hormozi offers, PAS, AIDA, variation generation, LATAM hooks
Synthetic Demand Validation (SDV) — predict purchase intent, price tolerance, and objections for ad creatives, offers, and landing copy by eliciting free-text reactions from a synthetic persona panel and mapping them to Likert distributions. Comparative ranking is the headline signal. Use for "will this convert", "which variant wins", "is $X the right price", "pre-test this creative/offer/copy before spend".
Cross-channel GTM orchestration — channel sequencing, attribution models, retargeting chains, multi-touch journeys
| name | campaign-optimization |
| description | Campaign optimization methodology — budget allocation, creative testing, scaling playbook, week-by-week ops |
Systematic methodology for optimizing paid acquisition campaigns from launch through scale. Covers the full lifecycle: learning phase management, creative testing, budget allocation, and scaling.
Meta's algorithm needs 50 optimization events per ad set per week to exit the learning phase. During days 1-3:
Minimum thresholds before making any decision:
| Metric | Minimum data needed |
|---|---|
| Impressions per ad | 1,000 |
| Clicks per ad | 50 |
| Conversions per ad set | 15 |
| Spend per ad set | 2x your target CPA |
Review performance but only make minor adjustments:
After 7 days with sufficient data:
| Metric | Formula | Good benchmark (EdTech LATAM) |
|---|---|---|
| CPA (Cost Per Acquisition) | Spend / Conversions | $5-25 (lead), $50-200 (paid user) |
| ROAS (Return on Ad Spend) | Revenue / Spend | 3x+ for sustainable growth |
| CAC (Customer Acquisition Cost) | Total marketing cost / New customers | < 1/3 of LTV |
| Metric | What it tells you | Alarm threshold |
|---|---|---|
| CTR (Click-Through Rate) | Creative resonance | < 1% = creative problem |
| CPC (Cost Per Click) | Audience + creative quality | > $2 LATAM = targeting issue |
| CVR (Conversion Rate) | Landing page effectiveness | < 2% = page problem |
| CPM (Cost Per 1000 Impressions) | Auction competitiveness | > $15 LATAM = audience too narrow |
| Frequency | Ad fatigue signal | > 3 in 7 days = rotate creative |
| Hook Rate (video) | First 3s retention | < 25% = weak hook |
| Hold Rate (video) | Full video views / impressions | < 5% = content issue |
| Thumb-Stop Ratio | 3s views / impressions | < 20% = visual not stopping scroll |
Test in this order (highest impact first):
Never kill a creative without sufficient data:
| Daily budget per ad set | Min days before decision | Min conversions |
|---|---|---|
| $20-50 | 5-7 days | 15 per variant |
| $50-100 | 3-5 days | 25 per variant |
| $100+ | 2-3 days | 50 per variant |
A creative is a "winner" when:
| Allocation | Purpose | What goes here |
|---|---|---|
| 70% | Proven performers | Ad sets with CPA below target, scaling |
| 20% | Testing | New audiences, new creative, new angles |
| 10% | Moonshots | Completely new approaches, new platforms |
Goal: Find winning creative + audience combinations.
Goal: Optimize and expand what works.
Goal: Maximize volume while maintaining efficiency.
Goal: Dominate your category.
The single most important optimization rule in the Andromeda era: when profitable, DO NOTHING.
If CPA <= 1.2x target for 5 straight days at $500+/day spend: lock the ad set for 14 days.
During the lock period:
Andromeda's pattern recognition builds a compounding model of your ideal customer. Every edit -- even a "minor" one -- resets this model. Frequent edits interrupt the algorithm's ability to identify and expand high-value audience pockets.
The top $100M+/year advertisers are patient. They let winning ad sets run undisturbed while concentrating all iteration in separate testing campaigns. The scaling campaign is a "set it and forget it" machine.
Only break the 14-day lock if:
Meta introduced two new delivery statuses that provide early warning of creative exhaustion:
Add Event Match Quality to your weekly optimization audit alongside CPA, ROAS, and creative health.
external_id, em, ph, fbp/fbc, ct, client_user_agent are sent with every eventIn the post-attribution-upheaval world (March 2026), platform-reported ROAS is unreliable. Use blended MER as your source of truth.
MER = Total Ad Spend (all platforms) / Total Revenue (all sources)
| MER | Interpretation |
|---|---|
| < 0.15 | Excellent -- spending $0.15 to make $1.00 |
| 0.15 - 0.25 | Good -- sustainable for most margins |
| 0.25 - 0.40 | Acceptable -- watch unit economics closely |
| > 0.40 | Danger -- spending $0.40+ to make $1.00, likely unprofitable |
Andromeda-era budget scaling is more conservative than pre-Andromeda.
Day 1: CPA at target, spending $100/day
Day 5: CPA still at target → increase to $120/day (+20%)
Day 9: CPA still at target → increase to $144/day (+20%)
Day 13: CPA still at target → increase to $173/day (+20%)
Day 17: CPA still at target → increase to $207/day (+20%)
At this rate, you 2x budget in ~3 weeks without ever triggering a learning phase reset. Patience compounds.