使い方
- インストールコマンドをコピーし、自分の環境へ導入する前にソースを確認します。
- 必要なツール、認証情報、対応環境を原文で確認します。
- 入力と成果物を指定し、出力を資料と照合します。
npx skills add https://github.com/TerminalSkills/skills --skill media-buying開始前の確認
- 導入にはモデル契約、API枠、ファイルへのアクセス許可は含まれません。
- スター数や導入数は活動の指標であり、独立した品質・安全性試験ではありません。
見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。
Media Buying
Overview
Plan and execute paid media across programmatic, direct, and social channels. Optimize spend allocation, audience targeting, and cross-channel performance.
Instructions
Budget allocation framework
Allocate budget across channels based on funnel stage. The percentages and CPM/CPC/CPA ranges below are planning heuristics, not sourced benchmarks; replace them with your own account data.
AWARENESS (top of funnel) — 30-40% of budget (heuristic)
├── Programmatic display (CPM $2-8)
├── YouTube/CTV (CPM $10-25)
├── Podcast sponsorships (CPM $15-30)
└── Influencer partnerships
CONSIDERATION (middle) — 30-40%
├── Paid social (Meta, TikTok, LinkedIn)
├── Sponsored content / native ads
├── Retargeting display (CPM $3-12)
└── Search non-brand keywords
CONVERSION (bottom) — 20-30%
├── Search brand keywords (highest ROAS)
├── Retargeting (cart abandoners, high-intent)
├── Email/SMS (near-zero marginal cost)
└── Affiliate partnerships (CPA-based)Media plan template (all figures illustrative)
## Media Plan — Q2 2026
**Objective**: Generate 500 qualified leads at <$80 CAC
**Total Budget**: $40,000/month
| Channel | Monthly Budget | Model | Target CPA | Expected Volume |
|---------|---------------|-------|------------|-----------------|
| Google Search (brand) | $5,000 | CPC | $25 | 200 leads |
| Google Search (non-brand) | $8,000 | CPC | $65 | 123 leads |
| Meta Ads | $12,000 | CPA | $90 | 133 leads |
| LinkedIn Ads | $8,000 | CPC | $120 | 67 leads |
| Programmatic Display | $4,000 | CPM | N/A | 500K impressions |
| Content syndication | $3,000 | CPL | $75 | 40 leads |Programmatic advertising (RTB)
1. User visits a webpage with ad space
2. Publisher's SSP sends bid request to ad exchange (typical RTB timing; illustrative)
3. Ad exchange broadcasts to connected DSPs
4. DSPs evaluate user data + campaign rules → submit bid (illustrative timing)
5. Highest bidder wins, ad is served (~100ms total)
Key players:
- DSP: Where advertisers buy (DV360, The Trade Desk, Amazon DSP)
- SSP: Where publishers sell (Google Ad Manager, Magnite)
- DMP/CDP: Audience data (Lotame, Adobe Audience Manager)DSP campaign structure: Advertiser → Campaign → Insertion Order (by objective) → Line Item (by audience) → Creatives.
Targeting layers
AUDIENCE: First-party data, third-party (DMP), contextual, lookalike, intent signals
INVENTORY: Domain allowlists/blocklists, app vs web, viewability (MRC standard: 50% of pixels for 1 continuous second for display, 30% for large display ads of 242,500+ pixels, 50% for 2 continuous seconds for video), ad position
ENVIRONMENTAL: Geography, device, dayparting, frequency caps, browser/OSBuying models
All ranges below are illustrative, not benchmarks.
CPM (Cost Per 1000 impressions): Awareness/reach. Range: $2-50
CPC (Cost Per Click): Consideration/traffic. Range: $0.50-15
CPA (Cost Per Acquisition): Performance/conversion. Range: $10-500
CPV (Cost Per View): Video campaigns. Range: $0.01-0.10
CPL (Cost Per Lead): B2B lead gen. Range: $20-200Brand safety
Pre-bid: Domain allow/blocklists, category exclusions, keyword blocklists,
brand safety vendors (Integral Ad Science, DoubleVerify)
Post-bid: Viewability measurement (MRC standard, see above), IVT detection,
brand suitability scoring, placement reportsCross-channel optimization
Attribution windows: each platform sets its own click and view windows, and they change over time. Read the current setting in each platform's ad settings before you compare CPA across channels. Meta's Ads Manager offers several click and view windows; do not assume a fixed default. Do not mix windows in one blended CPA.
Budget reallocation: Review weekly. A heuristic is to move 10-20% from high-CPA to low-CPA channels, wait about a week, and remeasure. If CPA increases after budget increase, you've hit diminishing returns — pull back. Optimize for marginal CPA, not average CPA.
Negotiating direct deals
For premium placements (homepage takeovers, newsletter sponsorships, podcast ads): request the rate card and negotiate a discount (the size depends on the deal; no benchmark is implied), ask for added value (bonus impressions, social posts), request performance guarantees and historical data, negotiate cancellation terms, get makegood policy in writing.
Examples
Create a programmatic media plan
We're launching a B2B SaaS product targeting CTOs and VP Engineering at companies with 100-1000 employees. Budget is $25,000/month for programmatic display and video. Create a full media plan using DV360 — campaign structure, audience segments (first-party site visitors + third-party intent data), creative specs, frequency caps, and brand safety settings. Include a 4-week optimization timeline.Optimize cross-channel budget allocation
We run ads across Google Search ($15K/mo), Meta ($20K/mo), LinkedIn ($10K/mo), and programmatic display ($5K/mo). Our blended CPA is $95 but varies wildly by channel. Analyze the marginal CPA for each channel at different spend levels and recommend a reallocation to reduce blended CPA by 20%. Include the data I need to collect and the rebalancing schedule.Set up brand safety for programmatic campaigns
We're a financial services company running programmatic display across 500+ publisher sites. Set up comprehensive brand safety — domain allowlist of premium financial and business publishers, category exclusions, keyword blocklists specific to our industry, and viewability thresholds. Include a weekly audit process to catch new placement issues.Guidelines
- Always set up brand safety (allowlists, blocklists, verification vendors) before launching programmatic campaigns
- Optimize for marginal CPA, not average CPA — the next dollar should go to the cheapest conversion
- Avoid large one-week budget moves (a heuristic cap of about 20% per channel per week) because sudden changes can disturb platform learning phases
- Use frequency caps to limit ad fatigue; set the cap from your own frequency-versus-conversion data (3-5 per day is a common starting heuristic, not a standard)
- Always request historical performance data and makegood policies before signing direct deals
- Review placement reports weekly to catch brand safety issues early
- Include viewability thresholds in all programmatic line items: the MRC standard is the minimum, and a stricter buyer threshold (e.g. 70%) is a choice you make, not a standard
