AMAZON · MCP data analysis
Xiyou MCP keyword analysis guide: shortlists, trends and matrices
Our Xiyou MCP library has 22 templates across nine operations scenarios. Start keyword research with template 07, assess individual competition with 09, and compare candidates with 16.
What the three templates address
| Template | Suitable question | Expected output |
|---|---|---|
| 07 · Keyword shortlist | Initial keyword strategy for a product or ASIN | Reverse-search ASIN keywords and combine searches, competition, suggested bids and ABA trends into core, long-tail and watchlist tiers. |
| 09 · Keyword first-page difficulty | Understand the competitive threshold for page one | Compare competing ASINs' ratings, reviews and price bands, alongside monthly keyword competition history. |
| 16 · Keyword comparison matrix | Narrow an existing advertising candidate list | Retrieve basics for up to 20 candidates, check weekly ABA trends for core terms and create a comparable matrix. |
These are outputs the prompts ask AI to organize. Completeness depends on actual tool responses. The template page only selects marketplaces and copies text; it does not call models, query data or create campaigns.
Prepare marketplace, products and comparison conditions
- Connection: Connect Xiyou MCP in your own AI client and confirm available tools and credits. Use the current Xiyou console for access and package rules.
- Marketplace: For example US, DE or JP. Analysis and returned data must use the same marketplace; do not mix search volumes or currencies across markets.
- Products and ASINs: Confirm product name, target ASIN and any competitor ASINs. Amazon links are also accepted in templates involving ASINs.
- Candidates and business conditions: Template 16 needs up to 20 terms, a core-term count and order value placeholders. For advertising decisions, also prepare your costs, margins and conversion data.
Do not treat placeholders or demo ASINs as real targets. Check every bracket for marketplace, product, dates, word count and currency before sending.
Steps for one analysis
- Open the library and select a marketplace. Keywords and traffic section contains templates 07, 08 and 09; template 16 is in the Batch and matrix section . Country buttons replace marketplace placeholders except template 17's multi-market text.
- Copy and replace inputs. Replace product, ASIN, keyword and order-value brackets with confirmed inputs. Keep clear periods; template 07 reverse-searches ASIN keywords from the last seven days.
- Send to an MCP-connected AI assistant. The prompt states the business goal and the client routes available tools. Listed tool chains explain sources; they do not prove every client is connected.
- Review returned data before conclusions. Check actual tools, marketplace, dates, units and missing values. Explicitly mark unavailable metrics instead of inventing values from model knowledge.
- Narrow candidates and keep evidence. Create core, long-tail and watchlist tiers, then add trend/competition data for a few main terms. Save tool responses and tables for later review against real advertising performance.
Template 07's reference chain is get_asin_keywords → get_keyword_info → get_keyword_aba_trends ; template 09 also queries competing ASINs and monthly history. Tool names follow the original templates; use the currently exposed MCP tools and parameters.
What a keyword result should let you verify
- Basics for each keyword
- Associate every term with marketplace, period and tool source. Keep units and periods for search volume, ABA rank, competition, suggested bid and weekly trends; mark missing values separately.
- Tiering can be traced to data
- Explain relevance, demand and competition for core, long-tail and watchlist terms. High searches/low competition is a direction for screening; check meaning against selling points and use cases.
- Concrete competitors behind the threshold
- Page-one difficulty should identify competing ASINs, price bands, ratings and review thresholds, rather than only labeling a term easy or hard.
- Explainable ranking and exclusions
- Template 16 uses a suggested-bid threshold of 15% of order value, a heuristic from the original prompt. It does not prove break-even click costs and cannot replace actual conversion and margin calculations.
Targeting suggestions are analytical advice. Without ad reports, real CPC, conversion and profit data, they cannot establish profitability, suitable budgets or successful optimization. Record observations and business actions separately.
Limits to retain
- Original templates use Xiyou MCP parameter definitions dated 2026-08-27. Tools, marketplaces, historical coverage and fields may change. A prompt request does not guarantee an API can return it.
- This page does not promise real-time collection, full coverage or a specific quota. Credit is a quota unit, not a call count; official account pages determine current eligibility and quotas.
- Marketplace selection only replaces text. Check both the copied prompt and returned marketplace. Align periods, currency and category definitions when comparing markets.
- Third-party search and competition data is an analytical reference, not proof of causes for your own ASIN's sales changes. For declines, combine template 08 with your own backend evidence.
- A candidate table is not an executed ad strategy. Remove store details, costs, ad reports and other private business data before sharing. This guide includes no private merchant/product case.
Template sources and further reading
Based on this site's existing prompts, tool chains and instructions, helping you understand inputs/outputs before copying full prompts.