SuperAPI AI Directory
Partner中文AI newsFavorites
Skillszach-search-term-analyzer

zach-search-term-analyzer

General & other

View repository

Updated 10-16 09:56

About this skill

Amazon Brand Analytics Search Term Analysis (v3 Scenario-Based Framework)

Analyze the Brand Analytics Top Search Terms report (any category), route scenarios based on "whether your own ASIN is/is not in the TOP3 clicks," and output market structure judgments and practical recommendations.

Data Source (⚠️ Important Distinction)

This tool analyzes the Brand Analytics → Top Search Terms report,not the PPC Search Term report; if the file fields do not contain the required fields below, stop first and confirm the report type.

Brand Analytics Top Search Terms PPC Search Term Report
Source Seller Central → Brand Analytics Ad console → SP/SB Campaign
Data Scope All-platform search terms (can be checked even if unrelated to your product) Only search terms related to your ad triggers
Core Metrics Search frequency rank, click share, conversion share Impressions、Clicks、Spend、Sales、ACOS
Applicable Scenarios Market analysis, product selection positioning, Listing keyword strategy, monitoring your own term share Ad optimization, bid adjustment, negative keyword management

⚠️ Using Brand Analytics requires completing Brand Registry. For the PPC Search Term report, please use zach-search-term-report-analyzer。

Report Definitions Quick Reference (Must-Read Before Interpretation)

Definition Content
Search Frequency Rank All-platformRelative rankingNot search volume; #1 and #2 can differ by orders of magnitude; not comparable across categories/sites
TOP3 Products Under that term,most clickedthe 3 ASINs (including ad placement clicks) — they are "click kings," not "conversion kings"; the ASIN with the highest conversion may not be in the table
Click Share / Conversion Share Independent metrics: buyers can click A and buy B, so TOP3 conversion share being 0/missing is a common real phenomenon (conversions spread to the long tail or are anonymized), not a data error
Conversion Coefficient Conversion share ÷ click share = that ASIN's conversion rate ÷ that term's average conversion rate; ≥1 = Closer, <1 = traffic magnet
Granularity and Lag Daily/weekly/monthly/quarterly selectable; in practice,weekly reportsare primary; data lag is about T+3~T+7; low-frequency terms are anonymized and not displayed
Positioning Boundary ABA represents popular traffic, not necessarily the full picture of market demand

Data Integrity Rules (⭐ Highest Priority)

  1. Never fabricate data: all analysis data comes from the raw report files provided by the user; if there is no data, say "not available"
  2. Missing ≠ Zero: terms with missing/all-zero conversion share are listed separately in B0 "insufficient data," do not participate in quadrant determination, and must not be judged as "poor conversion"
  3. Distinguish facts from inferences: data conclusions are marked "📊 Data Fact," strategy recommendations are marked "💡 Analytical Inference"
  4. Ranking ≠ Search Volume: Search frequency ranking only indicates relative position
  5. Timeliness: It is recommended to use data from the most recent 3-6 months; for strongly seasonal products, refer to the same period of the previous peak season, and analyze different periods separately

Scenario Framework (v3 Core)

Scenario Routing

自家 ASIN ∈ 该词点击 TOP3 ？
├─ 是 → 场景 A：存量经营（看自家行 + 同词竞品行）
├─ 否 → 场景 B：市场进入（看 TOP3 合计的市场结构）
└─ 未提供自家 ASIN → 全部场景 B，报告显著标注"仅市场视角"

Scenario A: You are in the TOP3 (Defense/Harvest Perspective)

Status Determination (Default Threshold) Action
A1 Harvest Term Your conversion coefficient ≥ 1.0 Defend + scale up: increase budget, defensive placement, watch for competitor entry
A2 Leaky Term Your click share − conversion share > 4pt Fix the detail page before raising bids: price/reviews/main image/A+, use SQP to locate funnel bottlenecks
A4 Share Warning Was once in TOP3, dropped out in the latest period (or click share has been declining continuously) Investigate ranking/price/stockouts/new competitors
A5 Balanced Term Between the two Maintain the status quo, track weekly

Additional signal: Competitors within the TOP3 for the same term have a conversion coefficient <1 → Traffic interception opportunity(SP/SD targeting + price suppression).

Scenario B: You are not in the TOP3 (Attack/Entry Perspective)

Two axes:Click concentration = SUM(TOP3 click share);Top satisfaction = SUM(conversion share) ÷ SUM(click share):

Quadrant Determination (Default Threshold) Structural Interpretation Action
B1 Opportunity Term Concentration <50% and satisfaction <1.0 Clicks are dispersed and the top players cannot capture demand; conversions flow outside the TOP3 Highest priority: If the product can capture the intent, prioritize placement + add to Listing; first verify TOP3 attribute consistency
B2 Regular Competitive Term Concentration <50% and satisfaction ≥1.0 The pie is not consolidated, and top players have healthy conversion Can enter: regular differentiation + normal bid testing
B3 Top Satisfaction Term Concentration ≥50% and satisfaction ≥1.0 Top players monopolize and efficiently satisfy the market Avoid their edge: do not bid head-on; find adjacent long-tail entry points
B4 Ambush Term Concentration ≥50% and satisfaction <1.0 Top players dominate clicks, but buyers purchased from others Ambush: competitor targeting + price suppression, steal orders with conversion power, do not fight a click war
B0 Insufficient Data Conversion data missing/dispersed/desensitized Cannot determine top satisfaction Do not determine; switch to SQP/advertising data for verification

Additional signals:Monopoly red line(click or conversion concentration ≥70%);Blue ocean signal(SFR ≥200,000 and TOP3 share ≥60%, an undeveloped niche).

SFR Five-Level Tiering (Budget Scale Label, Not Involved in Quadrant Determination)

T1 Core Big Terms (≤10,000, for ranking) | T2 Sub-core (10,000-20,000) | T3 Mid-volume Terms (20,000-50,000) | T4 Mid-long-tail (50,000-100,000, ROI-oriented) | T5 Long-tail (>100,000, chase ACOS)

All thresholds are default values derived from multi-source practical experience and can be calibrated by category via script parameters; for judgment criteria and sources, see references/关键词分类逻辑说明.md。

Execution Steps

  1. Identify the CSV file - Locate the search term report in the current directory or a specified directory
  2. Determine your own ASIN list (scenario routing) - Ask the user to provide their own ASIN list; if it cannot be obtained, clearly state "analyze everything from the Scenario B market perspective" before continuing, and do not silently skip
  3. Run the single analysis engine - Use the script to perform the complete analysis; temporary handwritten substitutes are prohibited: 
    python3 skills/zach-search-term-analyzer/scripts/analyze_search_terms.py <报告目录> \
  --own-asins "B0XXXXXXXX,B0YYYYYYYY"   # 或 --own-asins-file <清单文件>
    • Cross-category can add --category-lexicon <品类词表.json>（audience_words / scene_words / attribute_words three optional string arrays)
    • Category calibration thresholds:--click-conc-concentrated、--monopoly-redline、--coefficient-threshold、--own-gap-points、--sfr-tiers , etc.
  4. (Optional) Semantic annotation pass - The script's keyword tags are only a preliminary screening by a surface-form dictionary; the AI assistant can add buyer intent annotations for the TOP 200 search terms (who is buying / in what scenario / for what task). Annotations are incremental additions only: they do not filter out, reorder, or modify the script's quantitative classification
  5. Interpret the report - Provide placement pool allocation and Listing layout suggestions according to the scenario framework; ⚠️ negative keywords are not automatically generated by this report—negative keyword decisions are driven by your own PPC search term report

Output Files

analysis_reports/ directory:

File Content
01_品牌竞争分析.csv/.md Brand TOP3 position ranking frequency (⚠️ not market sales share)
02_类别趋势分析.csv/.md Category distribution and cross-category opportunities
03_搜索词热度分析.csv/.md ⭐⭐ Core output of scenario-based classification: scenario/status/concentration/satisfaction/red line/blue ocean/own share
04_词频统计分析.xlsx/.md ⭐ Single words + two-word phrases (deduplicated counts; phrases only take real adjacent combinations)
05_竞争格局分析.csv/.md Click TOP3 product level (own ★ marked)
06_成交系数分析.csv/.md Per-ASIN conversion coefficient: own self-diagnosis + traffic interception targets + reverse learning targets
07_时间序列分析.csv/.md Period overview + core keyword trends (when multiple periods, includes 核心关键词趋势.csv/.png）

Report Format Requirements

  • CSV file (UTF-8 encoding), supports multi-period reports; duplicate (search term, report date) across files is automatically deduplicated; share fields are compatible with numeric and 12.34% strings
  • Required fields: search frequency rank, search term, report date, top-clicked products #1/#2/#3 (ASIN/click share/conversion share); when brand/category fields are missing, an empty placeholder report is produced for the corresponding dimension (some ERP exports such as Lingxing do not include brand fields)

Practical Suggestion Examples

🧭 场景 A（自家 ASIN 已提供）：
🛡️ wireless speaker waterproof（A1收割词，自家系数1.3）→ 加预算防守
🔧 bluetooth speaker（A2漏水词，点击12.8% vs 转化6.1%）→ 先修详情页再抬价
🎯 同词竞品 B0XXXXXXXX 系数 0.4 → SP 商品定向截流

🧭 场景 B（市场进入）：
🌊 speaker with lights（B1机会词，集中度28%、满足度0.6）→ 属性一致则优先投放+进标题
🥷 party speaker（B4伏击词，集中度61%、满足度0.5）→ 竞品定向偷单，不打点击战
⛔ jbl speaker（B3头部满足词+垄断红线）→ 避其锋芒找长尾
❓ karaoke machine kids（B0数据不足）→ 转 SQP/广告数据验证

Local References

  • references/关键词分类逻辑说明.md — complete judgment criteria, threshold basis, and sources for the scenario framework
  • references/字段说明.md — report field definitions and pitfalls
  • references/分析角度和方法.md — 7-dimension analysis method and practical applications
  • scripts/analyze_search_terms.py — single analysis engine (v3.0)

Dependencies

  • Python 3.8+; dependency check: 
    python3 -c "import pandas, numpy, matplotlib, seaborn, openpyxl"
  • If missing:pip3 install -r skills/zach-search-term-analyzer/scripts/requirements.txt

Risks and Boundaries

  • risk-level: low — pure local analysis, does not involve any account operations or data writes; your own ASIN list is only used for local routing calculations
  • All judgments in this report arepriors: B1's "outflow can be captured" and A2's "leakage cause" must ultimately be verified with SQP (own funnel share) and PPC data (actual CVR/ACOS)

After Completion

Report completion status: DONE / DONE_WITH_CONCERNS / BLOCKED / NEEDS_CONTEXT