Naver Product Shopping Scraper

Naver Product Shopping Scraper searches Naver Shopping and returns one normalized product comparison record per result. It is for Korean ecommerce teams, catalog analysts, and price-monitoring workflows that need current product titles, prices, sellers, brands, categories, ratings, review counts, product links, and collection timestamps.

Use it for Naver Shopping search-result research. Do not use it for full SmartStore or Brand Store catalogs, per-SKU product details, or review-text collection; those are different jobs with different data shapes.

What it extracts

Each dataset item represents a single public Naver Shopping search result. The actor normalizes productId, title, productUrl, priceKrw, originalPriceKrw, discountPercent, seller and brand data, category labels, rating, review count, sourceUrl, and scrapedAt. It uses explicit null values when Naver does not expose a field.

When to use it

Use this actor to compare Naver Shopping offers for a keyword, monitor broad product coverage, or prepare a price-intelligence dataset for another workflow. Start with a small maxResults value to confirm the current result shape and cost.

Do not use it when you already have Naver SmartStore or Brand Store product URLs and need variant-level stock, long-form descriptions, or legal disclosures. It also does not collect review text or bypass login-only content.

Input

{
  "searchQueries": ["wireless earbuds", "무선 이어폰"],
  "maxResults": 20
}

• searchQueries accepts up to 20 product phrases.
• maxResults is both the dataset and product-result billing cap, from 1 to 200.
• The actor uses persistent Korean residential browser sessions internally, with an owner-configured residential fallback for Naver's IP-reputation blocks; no Naver API key or cookie is accepted as input.

Output

{
  "productId": "1234567890",
  "title": "Wireless Bluetooth Earbuds",
  "productUrl": "https://search.shopping.naver.com/catalog/1234567890",
  "priceKrw": 59900,
  "sellerName": "Example Mall",
  "brand": "Example Brand",
  "rating": 4.8,
  "reviewCount": 123,
  "sourceUrl": "https://search.shopping.naver.com/search/all?query=wireless%20earbuds",
  "scrapedAt": "2026-07-14T12:00:00.000Z"
}

The default key-value store also contains OUTPUT for the compact terminal outcome and RUN_SUMMARY for safe diagnostic, coverage, and billing counters. Valid queries with no matches return VALID_EMPTY; malformed input returns INVALID_INPUT; a blocked upstream route with no usable result fails honestly as UPSTREAM_FAILED.

Pricing

This actor uses Pay per event + usage: actor start is $0.00005 per run and each validated product result is $0.06. The price supports the reliable managed Korean search route; Apify platform usage is charged separately. The managed route invokes a separate Apify Actor under the same account, so its own provider events can also appear in account billing. A 20-result run has at most $1.20005 in this actor's events, plus platform usage and any managed-route provider events.

API and MCP use

Run the actor through the Apify Console, API, schedules, or Apify MCP. For an agent, supply a concise product phrase and a bounded maxResults, then read the dataset through the run’s default dataset endpoint. Each item includes sourceUrl and scrapedAt so downstream steps can retain provenance.

Limits and reliability

Naver may vary search-result fields by query, region, and current site behavior. The actor uses a browser response collector with a DOM fallback, validates every completed record before it is persisted, and keeps a single browser session per request. A Korean residential proxy is required because Naver can reject direct non-Korean HTTP traffic.

Legal and responsible use

Use this actor only for data you are authorized to collect and process, and comply with applicable law, Naver’s rules, and downstream data-use obligations. Do not use it to collect login-only data, circumvent access controls, or process personal data without a lawful basis.

Feedback

If a specific search phrase starts returning an unexpected shape, include the run ID, query, and a redacted example of the affected record when reporting the issue. That lets us distinguish a query-specific empty result from a site-shape change.