Find Shopify traffic leaders by category.

Cost guard: before each paid Spytrend call, run the free `estimate_cost` with the same tool and arguments, estimate the cost and stay within the user's stated budget; if the user has not stated a budget, tell them the estimate before the first paid call. If pricing consent is required, show the returned terms and stop until the customer explicitly agrees; never accept on the customer's behalf. If a required call does not fit the stated budget, make no paid call and return `Budget limit prevents required research`; omit an optional call that does not fit and disclose reduced coverage. When opening ads with `get_ad`, skip multilang ads (`is_multilang`) and never call `get_media` after `get_ad`. Retry once only when `retryable=true`, after `retry_after_seconds`. Read `params.ignored` after every search; an ignored required filter never counts as applied. If a search reports `pagination.total_status=restricted` (a free-plan archive window; `meta.plan_window_cutoff` names the boundary), tell the user the result covers only ads that started on or before that date, and never report a zero from it as "no such ads exist". Cost notices are not part of the final customer report.
Use the connected Spytrend MCP to prepare a customer-facing traffic benchmark of category-relevant Shopify stores for an independently selected market context.

Execute every research stage in this prompt in one turn. Apart from mandatory pre-call cost notices, do not emit intermediate step responses and do not stop after validation. Stop early only for invalid input, unavailable required evidence or unresolved required identity. Return only the final customer report.

This workflow helps a customer choose Shopify storefronts to benchmark for navigation, merchandising and returned acquisition mix. It ranks verified stores by returned global estimated monthly visits within a bounded inspected sample. It is not a complete market ranking, country-specific traffic ranking, competitor-ad workflow, growth workflow or measured analytics report.

Configuration:
- Category label: beauty_and_personal_care
- Spytrend AI category: beauty_and_personal_care
- Market label: US
- Market ISO-2: US
- Platform: Shopify
- Minimum estimated visits: none (`none` means no threshold)
- Usable traffic quality: medium, high
- Discovery limit: 15
- Compact store inspection limit: 15
- Ranked result limit: 5

Input validation:
1. Continue only when the category and market are supported values from the page's category and market lists. Do not bind a category to a country. Accept no customer-written slug, query, subcategory, synonym, product rule, brand or country. Reject unsupported values such as `apparel_and_accessories`. If invalid, return exactly `Invalid controlled input configuration` and make no MCP call.

Decision and evidence semantics:
2. The customer decision is: which category-relevant Shopify storefronts in the selected store-country context deserve closer benchmarking based on the largest usable global traffic estimates in this inspected sample?
3. The primary entity is one normalized store domain. Keep store, company, advertiser, page and webmaster entities separate.
4. US is discovery and returned store-country context. It does not prove headquarters, registration, shipping, customers, sales, traffic dominance or market share.
5. Treat monthly visits as a global third-party estimate unless a country-specific scope is explicitly returned. Never call the result `most visited in US`.

Research:
6. Use Spytrend read-only.
7. Run heavy calls strictly sequentially.
8. Run exactly one search_shops call with ai_category=beauty_and_personal_care, country=US, platform=Shopify, sort_by=visits, sort_order=desc, exclude_infrastructure=true, optional min_visits and limit=15.
9. Do not add advertising, growth, q, app, theme, plan, brand, generated synonym, subcategory or fallback-category filters.
10. Preserve returned order. Locally skip only invalid domains, infrastructure, explicit marketplaces/general stores and rows whose returned category explicitly contradicts beauty_and_personal_care. Do not skip unfamiliar brands or inconvenient estimates.
11. Use the returned discovery rows directly as the compact store records. Normalize and deduplicate exact registrable domains, preserve returned order and stop when five stores qualify as Ranked or the 15-row discovery sample is exhausted. Never call `get_shop` or repeat `search_shops(q=domain)` for a domain already returned by this controlled discovery.
12. Use the compact record for Shopify identity, category, returned store-country context, estimated visits, traffic quality and optional store signals. Do not run advertising enrichment in the base traffic-ranking workflow. Advertising is not an eligibility requirement and may be requested later as a separate, newly budgeted enrichment.
13. Do not call `search_ads`, `get_ad`, `get_advertiser`, `get_trends`, `search_webmasters` or other enrichment tools in the base run.

Qualification:
14. `Ranked benchmark store` requires explicit Shopify/domain evidence, compact category evidence directly aligned with beauty_and_personal_care, an eligible merchant entity, US in returned store-country context, positive global estimated monthly visits and traffic_quality medium or high. A confirmed active ad may strengthen category fit but is not required.
15. `Unranked reference` requires Shopify, direct category fit, eligible merchant and market context, but traffic quality or estimate is low/missing. Keep it outside the numbered ranking.
16. Exclude infrastructure, marketplaces, affiliates/catalogs, publishers, general stores, taxonomy-only matches, stores where the category is incidental, absent market context and unresolved domains. Do not invent category-specific product rules. Never exclude an otherwise qualified store solely because an active representative ad was absent, unavailable or not confirmed.

Ranking and interpretation:
17. Rank at most five qualified stores by returned global estimated monthly visits descending. Preserve ties; never invent a score.
18. Show traffic_quality beside every estimate. Describe change only as `estimated monthly traffic change` with its returned comparison period.
19. Compare source shares only when their scope is comparable. Describe detected apps/theme/plan as incomplete observed store signals.
20. `Why benchmark it` must identify one concrete storefront question: navigation/category structure, merchandising/offer presentation or returned acquisition-source mix. Do not claim the store is successful because of that feature.
21. Never infer measured analytics, popularity, performance, customers, orders, conversions, sales, revenue, profitability, ROAS or market share.

Failure handling:
22. Retry exactly once only when retryable=true with retry_after_seconds, after a fresh identical estimate. A failed discovery makes the run Partial. A completed zero-row route is valid evidence.
23. Complete means the required discovery completed with required filters applied, not complete market coverage. Do not broaden category or country to fill the result.

Customer presentation:
24. Start `# Shopify Traffic Benchmark: [Category name]`, writing the category and market in plain words (for example Beauty and Personal Care, United States), never as slugs or ISO codes.
25. Then show `**[N] ranked benchmark stores** · **[M] unranked references** · Shopify · US store-country context · **[Complete/Partial]**`.
26. Immediately state: `Ranked within the inspected sample by returned global estimated monthly visits. The estimates are not country-specific unless explicitly stated.`
27. With at least two Ranked stores, add `## Benchmark shortlist`: `**[rank]. [Store]** — Estimated [visits] monthly visits · [quality] quality · [domain]`.
28. For each Ranked store use one compact paragraph of at most 65 words containing the traffic estimate and quality, direct category fit, one useful returned traffic or storefront signal, exactly one Why benchmark it question, and the domain as an explicitly returned stable first-party `Evidence` link. Do not repeat a separate Website line. Never name the service in the customer answer.
29. Put low/missing-quality direct-fit stores under `## Unranked references`, one line per store. Never compare their scale.
30. With at least two Ranked stores, add `## Benchmark patterns` with exactly two short sample-specific comparisons about storefront specialization, navigation/merchandising or comparable returned source mix. Do not call them market trends or causes of traffic.
31. Under `## Research coverage`, use at most three bullets for date/status, returned/inspected/ranked counts and bounded-sample/global-estimate limitations.
32. For zero Ranked stores, show `## Result`, optional Unranked references and Research coverage. Do not relax inputs.
33. Use no tables, raw calls, JSON, IDs, quotas, rejected-row dump or audit details. Keep the complete answer within 560 words. Missing data is `Not returned`, never zero or absent.

End exactly:
`Traffic figures are third-party estimates for research and may be incomplete. They do not prove customers, orders, sales, revenue, profitability, ROAS, market share or complete market coverage.`

Only after an exact separate request `Show audit details`, use retained responses to provide exact calls, route outcomes, returned order, skip/profile decisions, counts and timing without rerunning Spytrend.