Global Price Arbitrage — Cross-Border Price Comparison & Landed Cost Calculator

Most price comparison tools tell you a product is "$20 cheaper on the UK site" and stop there. This Actor keeps going: it searches Amazon and eBay across US, UK, Germany, Canada, and Australia storefronts, converts every price to one currency using live exchange rates, and then adds an estimated landed cost — shipping plus import duty — so you see what the product actually costs once it reaches you, not just the sticker price on a foreign page.

That's the gap this Actor fills. Plenty of Actors compare domestic retailers (Amazon vs. Walmart vs. Target). None of them tell you whether it's actually cheaper to import the same product from a different country once shipping and duty are accounted for — a price that looks 15% cheaper abroad is often not cheaper at all after those costs, and this Actor is the one that does that math for you.

What you get, per product

• Multi-region search — the same product searched across Amazon and eBay's US, UK, Germany, Canada, and Australia storefronts (you choose which), crawled with bounded concurrency you control.
• Live currency conversion — every price normalized into one currency of your choice, using live exchange rates (with an approximate fallback if the rate feed is briefly unavailable, clearly flagged).
• Landed cost estimate — item price + an estimated shipping cost + an estimated import duty/tax for every non-domestic marketplace, so cross-border options are ranked on what they'd really cost you, not just the listed price.
• Ranked comparison table — the cheapest listing from each marketplace, ranked by landed cost, so the actual best deal is obvious at a glance.
• A data-confidence grade (A+ to F) — scraped marketplace pages sometimes block or return nothing; this tells you what fraction of your selected marketplaces actually returned data, so you know how much to trust the ranking, not just whether the run "succeeded."
• A severity tag — critical (no marketplace returned any data), warning (some marketplaces returned data but not all — the ranking may be missing a cheaper option), or ok (every selected marketplace returned data) — for instant triage.
• A one-line executive summary — the best price found, in plain English, folding in any price-drop alert and data-confidence caveat.
• Price-drop monitoring (optional) — the Actor remembers the best option from your last check and flags it when the best deal changes or the price moves, pushed to Slack or your own webhook.
• AI-written buy verdict (optional, bring your own key) — paste your own Anthropic or OpenAI API key and get one plain-English sentence recommending where to buy and how much you'd save, instead of just reading a table.

Why this instead of checking manually

	Manual tab-checking	Global Price Arbitrage
Coverage	Whichever storefronts you remember to open	Every marketplace you select, automatically
Math	You eyeball the sticker price	Live FX conversion + landed cost (shipping + duty) per option
Trust signal	You just hope the page loaded right	Explicit data-confidence grade and severity per query
Repeatability	Re-checking is manual, tedious	One click, or schedule it with price-drop alerts
Output	Browser tabs and mental math	Structured dataset, exportable to CSV/JSON/BI tools

Who this is for

• Shoppers comparing a specific product across countries before an expensive purchase (electronics, cameras, watches, collectibles)
• Resellers and arbitrage sellers checking whether cross-border sourcing is actually profitable after shipping and duty
• Anyone who's seen "it's cheaper on Amazon UK" and wants to know if that's still true once it lands in their country
• Deal-hunters who want price-drop alerts on a specific product across multiple regions, not just one

How to use it

1. Add one or more products to compare (product name or search term).
2. Pick which marketplaces to check — more marketplaces means a more complete comparison but a longer run.
3. Set the currency you want prices compared in and your destination country (this determines which marketplaces count as domestic vs. cross-border).
4. Optionally override the built-in shipping estimate or set an import duty percentage if you know your category's typical rate.
5. Optionally paste a Slack/webhook URL to get alerted when the best deal changes, and/or your own Anthropic or OpenAI API key for a written buy verdict.
6. Optionally adjust max concurrent marketplace requests — the default (3) is deliberately conservative since Amazon and eBay block aggressively; raise it if you're on residential proxies and want faster runs.
7. Run it once for a snapshot, or put it on an Apify Schedule to track a product's best cross-border price over time.

Example output (one item)

{
  "query": "Sony WH-1000XM5 headphones",
  "targetCurrency": "USD",
  "destinationCountry": "US",
  "fxRatesLive": true,
  "grade": "A+",
  "dataConfidence": 100,
  "severity": "ok",
  "summary": "Best price: Amazon Germany at 291.40 USD -- Best deal on Amazon Germany dropped, 299.10 -> 291.40 USD (-2.6%). (5/5 marketplaces returned data.)",
  "successfulMarketplaces": 5,
  "totalMarketplaces": 5,
  "bestOption": {
    "marketplace": "amazon_de",
    "marketplaceLabel": "Amazon Germany",
    "title": "Sony WH-1000XM5 Wireless Headphones",
    "url": "https://www.amazon.de/...",
    "itemPrice": 279.40,
    "estimatedShipping": 12,
    "estimatedDuty": 0,
    "landedCostEstimate": 291.40,
    "currency": "USD",
    "isDomestic": false
  },
  "ranked": [
    { "marketplaceLabel": "Amazon Germany", "landedCostEstimate": 291.40 },
    { "marketplaceLabel": "Amazon US", "landedCostEstimate": 329.99 },
    { "marketplaceLabel": "eBay US", "landedCostEstimate": 341.50 }
  ],
  "diffSummary": "Best deal on Amazon Germany dropped, 299.10 -> 291.40 USD (-2.6%).",
  "aiVerdict": "Buy from Amazon Germany — landed cost of $291 beats the US price by about 12% even after estimated shipping."
}

Pricing

This Actor uses pay-per-event pricing:

• Query compared — charged once per product query processed, regardless of how many marketplaces were searched (all marketplace scraping, currency conversion, and landed cost calculation is bundled into this one event).
• AI verdict — charged only when you've supplied your own API key and a verdict was successfully produced. No key, no charge.

Note: AI verdicts require your own Anthropic or OpenAI API key, and that provider bills you directly for the underlying LLM call — this Actor's "AI verdict" event only covers the orchestration. Residential proxy is used by default for reliable results — Amazon and eBay rate-limit or block non-residential IPs quickly — with automatic retries and session rotation if a marketplace serves a block/captcha page. Marketplace requests run with bounded concurrency (default 3, configurable) rather than one at a time, so multi-marketplace runs finish faster without hammering any single storefront.

FAQ

What do the "grade," "severity," and "data confidence" fields mean? Marketplace scraping isn't guaranteed to work every time — a storefront can block a request, change its markup, or return an empty page. Rather than silently ranking whatever came back, this Actor tells you how complete the picture is: dataConfidence is the percent of your selected marketplaces that returned at least one listing, grade turns that into a letter (A+ to F), and severity is critical if nothing came back at all, warning if some but not all marketplaces returned data (the ranking may be missing a cheaper option), or ok if every marketplace responded.

How accurate is the landed cost estimate? It's a planning estimate, not a customs broker's calculation. Shipping is a flat estimate you can override; duty is a simple percentage you set (default 0%, since actual rates depend on product category, your country's de minimis threshold, and personal exemptions). Use it to compare options directionally — always confirm the real total at checkout before buying internationally.

Why do the same marketplace searches sometimes return different products? Search results are matched by keyword like a normal shopper would search, not by a universal product ID. For best results, search with a specific model name/number rather than a generic description.

Does this add anything to a cart or place an order? No. This Actor only reads public search results pages — no login, no cart, no checkout. It's read-only, the same way a human browsing and comparing tabs would be.

Does price monitoring work across scheduled runs? Yes — the best-option baseline is stored in your own Apify account (a persistent key-value store), keyed by product + currency + destination + marketplace selection, so it persists between runs regardless of how you trigger them.

How does the AI buy verdict work, and is my API key safe? Paste your own Anthropic or OpenAI API key and pick the matching provider. The Actor sends the ranked landed-cost table (numbers only, no personal data) to your provider with a tight prompt asking for one recommendation sentence. Your key is read from that run's input, used for that one HTTP request, and discarded — never logged, stored, or reused across runs. If you leave this blank, everything else works the same, you just read the ranked table yourself instead of a one-sentence verdict.

Is this legal? Reading publicly available product listing pages is generally legal; this Actor does not create accounts, log in, or interact with any site beyond searching and reading. Import duty and customs rules vary by country and change over time — this tool's duty estimate is not legal, tax, or customs advice, and you're responsible for verifying actual costs before making a cross-border purchase.