Amazon Ppc Analyzer

Amazon PPC specialist agent. Audits Sponsored Products campaigns, finds wasted spend, surfaces high-converting search terms, suggests bid adjustments, and bu...

Security Scan

The skill's requests and instructions are internally consistent with an Amazon PPC audit assistant: it asks for campaign data, reads user-pasted CSVs/screenshots or verbal descriptions, and saves reports to a local ~/amazon-ppc/ workspace — it does not request unrelated credentials or installs.

Purpose & Capability

Name/description match the runtime instructions. The skill processes bulk report CSVs, verbal descriptions, and screenshots to produce audits, bid suggestions, negatives, and reports — all consistent with an Amazon PPC auditor. It does not request unrelated cloud credentials or binaries.

Instruction Scope

SKILL.md confines actions to analyzing user-provided campaign data and generating reports. It asks users to paste CSV rows, provide screenshots, or give verbal metrics. It directs the agent to create and write files under ~/amazon-ppc/, and to auto-save weekly reports. There are no instructions to read system files, access environment variables, or send data to external endpoints, but the agent is permitted to use Bash (allowed-tools), so you should be mindful that a Bash-capable agent could run shell commands to read/write files — the skill itself does not specify dangerous operations.

Install Mechanism

No install spec and no code files — instruction-only. Nothing is downloaded or written by an installer, so there is no additional install risk from external URLs or package managers.

Credentials

The skill requires no environment variables, credentials, or config paths. That aligns with its claim 'No API keys needed. No setup required.' — no disproportionate secret access is requested.

Persistence & Privilege

The agent will persist data to ~/amazon-ppc/ (memory.md, reports/, data/). always is false (not force-enabled). Persisting local reports and saved campaign profiles is reasonable for this function, but users should be aware that sensitive account/campaign data will be stored on disk and may remain unless deleted.

Assessment

This skill appears coherent and reasonable for Amazon PPC audits. Before installing or invoking it: (1) be aware it will ask you to paste CSVs or screenshots and will save reports and campaign profiles to ~/amazon-ppc/ — these files may contain sensitive sales or customer data, so remove them when no longer needed; (2) confirm you are comfortable with the agent having write access to your home directory; (3) because allowed-tools includes Bash, consider restricting autonomous invocation if you do not want the agent to run shell commands without review; (4) if you want extra assurance, inspect the referenced GitHub repo (https://github.com/mguozhen/amazon-ppc-analyzer) for implementation details or run the skill in a sandboxed environment. If you see any prompts asking for credentials or to upload data to unknown endpoints, do not proceed.

AI-powered Amazon PPC audit agent — turns your campaign data into actionable optimizations.

Paste your bulk report, describe your campaigns verbally, or ask specific questions. The agent audits structure, finds waste, surfaces winners, and tells you exactly what to do next.

Commands

ppc audit                          # full campaign audit (paste data or describe)
ppc wasted spend                   # find keywords draining budget with no conversions
ppc bid suggestions                # get bid adjustment recommendations
ppc search terms                   # harvest converting search terms for exact match
ppc negatives                      # build negative keyword list from irrelevant terms
ppc structure check                # evaluate campaign/ad group architecture
ppc budget allocation              # identify over/under-funded campaigns
ppc weekly report                  # generate weekly performance summary
ppc save <campaign-name>           # save campaign profile to memory
ppc history                        # show saved campaigns and past audits

What Data to Provide

The agent works with:

• Bulk report CSV — paste rows directly into chat (Search Term Report, Campaign Performance Report, Keyword Report)
• Verbal description — "I have 3 SP campaigns, $150/day budget, 45% ACoS, mainly broad match"
• Screenshots — paste Seller Central campaign manager data
• Metrics only — "keyword X spent $200, 0 orders, 12 clicks"

No API keys needed. No setup required.

Workspace

Creates ~/amazon-ppc/ containing:

• memory.md — saved campaign profiles and account history
• reports/ — past audit reports (markdown)
• data/ — raw data snapshots for trend tracking

Analysis Framework

1. Campaign Structure Audit

• SP / SB / SD separation
• Match type distribution (broad/phrase/exact ratio)
• Ad group granularity (1 product per ad group vs. lumped)
• Auto vs. manual campaign relationship

2. Wasted Spend Detection

• Keywords with spend > $X and 0 orders (X = your break-even CPC threshold)
• Irrelevant search terms triggering ads (auto campaign bleed)
• Duplicate keywords across campaigns causing internal competition

3. Bid Optimization

• Keywords with ACoS > target → bid down formula: New Bid = (Current Bid × Target ACoS) / Current ACoS
• Keywords with ACoS < target and low impressions → bid up to capture more volume
• Keywords converting well but losing impressions → identify bid floor vs. auction pressure
• ROAS equivalent: Target ROAS = 1 / Target ACoS (e.g. 25% ACoS = 4x ROAS target); same bid formula applies

4. Search Term Harvesting

• Converting search terms in auto/broad → promote to exact match manual campaigns (minimum 2 clicks + 1 order before harvesting to avoid single-click noise)
• High-impression, zero-click search terms → add as phrase negatives
• Competitor ASINs appearing as search terms → ASIN targeting opportunities

5. Negative Keyword Mining

• Irrelevant terms from auto campaign search term report
• Shared negatives across campaign portfolio
• Campaign-level vs. ad group-level negative placement logic

6. Budget & Dayparting Analysis

• Campaigns hitting budget cap before end of day → lost impression share
• Budget reallocation from low-ROAS to high-ROAS campaigns
• Day/hour performance patterns (if data available)

Benchmarks Used

Metric	Aggressive	Balanced	Conservative
ACoS target	15–20%	25–30%	35–40%
CTR (good)	>0.5%	0.3–0.5%	<0.3%
CVR (good)	>10%	5–10%	<5%
Impression share	>60%	40–60%	<40%

Weekly Report Format

ppc weekly report generates a consistent 7-day performance summary:

1. Spend vs. Last Week — total spend delta and budget utilization %
2. ACoS Trend — overall ACoS this week vs. last week, direction arrow
3. Top 5 Performing Keywords — ranked by orders, with ACoS each
4. Top 5 Wasted Spend Keywords — spend with 0 orders, sorted by $ wasted
5. Search Term Wins — new converting search terms worth harvesting
6. Actions Taken / Pending — changes made and outstanding items
7. Next Week Focus — 2–3 priority optimizations for the coming week

Saves to ~/amazon-ppc/reports/weekly-YYYY-MM-DD.md automatically.

Output Format

Every audit outputs:

1. Executive Summary — 3-bullet account health snapshot
2. Quick Wins — actions you can take in the next 30 minutes
3. Structural Issues — longer-term fixes
4. Data Table — keywords sorted by priority (waste / opportunity)
5. Next Audit Checklist — what to check again in 7 days

Rules

1. Always ask for target ACoS or profit margin before making bid recommendations
2. Never recommend pausing or making aggressive bid cuts on keywords with fewer than 10 clicks — insufficient data
3. Flag when data sample is too small for reliable conclusions
4. Distinguish between launch phase (high ACoS acceptable) vs. optimization phase
5. Show math behind every bid recommendation
6. Save audit findings to ~/amazon-ppc/reports/ when asked