Naver Local Businesses Scraper

Refresh datasets of Naver local businesses from exact Korean Naver Map search URLs. Get structured business identities, categories, public contact fields, addresses, coordinates, review counts and booking flags without collecting review text.

This is a first-page snapshot tool for Korean local-market research, agency directory refreshes and local SEO analysis. It does not promise exhaustive geographic coverage.

What does it do?

Supply the exact Map search URL you want to snapshot. The Actor opens an anonymous Korean browser session with a lightweight search client, executes Naver's public session script, captures the public business search response and writes one row per unique place ID. It does not render the full interactive map. It processes URLs sequentially and stops at your global record limit.

Who is it for?

Korean local SEO analysts, agencies refreshing business directories and researchers comparing local business listings can use these snapshots. It is not an exhaustive lead database.

Why use it?

• Keep stable place IDs for joining successive snapshots.
• Preserve Korean names and source category labels without translation loss.
• Get contacts when already exposed in search, without expensive detail enrichment.
• Keep provenance: each row includes the originating URL, decoded query and extraction timestamp.
• Avoid mixing business records with reviews, photos or menu items.

Getting started

1. Open Naver Map and search using a Korean location and business keyword, such as 서울 카페.
2. Copy its https://map.naver.com/p/search/... URL.
3. Add it to searchUrls and choose maxItems.
4. Run the Actor and open the Businesses dataset.
5. Download JSON, CSV or Excel using Apify's dataset export tools.

{
  "searchUrls": ["https://map.naver.com/p/search/%EC%84%9C%EC%9A%B8%20%EC%B9%B4%ED%8E%98?c=126.97825,37.566551,13,0,0,0,dh"],
  "maxItems": 20
}

Inputs

Field	Meaning
searchUrls	Required array of 1–10 exact HTTPS Naver Map search URLs. Each path must contain Korean text. Only the optional c viewport parameter is supported.
maxItems	Global maximum unique businesses, default 20, range 1–200. Zero and unlimited are unsupported.

Duplicate URLs are processed once. Businesses are deduplicated across searches by place ID; the first occurrence supplies provenance. Earlier URLs can consume the entire limit. Only the first result page from each search is collected, so increasing the limit does not enable pagination.

The URL query is sent unchanged to Naver. Relevance, synonyms, proximity and ordering are controlled by Naver, not an exact business-name matcher. A location keyword or viewport is not a guaranteed geographic boundary; validate addresses before using results as an area census.

Extracted data

Fields	Description
placeId, name	Stable Naver identity and public name.
categories	Source category hierarchy, an array of strings.
telephone	Public business telephone, possibly a Naver forwarding number.
homepage	Homepage or social URL shown in search.
address, roadAddress	Lot-based and road-name addresses.
latitude, longitude	Numeric coordinates.
reviewCount, placeReviewCount	Two source counters; no individual reviews are extracted.
hasNaverBooking	Booking flag exposed in the listing, not live inventory.
placeUrl	Canonical mobile Place home URL.
sourceUrl, query, scrapedAt	Search provenance and UTC extraction timestamp.

Missing optional text and numeric values are null; categories may be empty. Counters are source values and need not agree. Missing booking flags normalize to false; this is not proof that a business never accepts reservations.

Output example

A representative public business observed during testing:

{
  "placeId": "1789177874",
  "name": "신이도가",
  "categories": ["음식점", "카페,디저트"],
  "telephone": "0507-1411-3461",
  "homepage": "https://www.instagram.com/shinleedoga",
  "address": "서울특별시 마포구 서교동 367-31 1층 신이도가",
  "roadAddress": "서울특별시 마포구 잔다리로 20-12 1층 신이도가",
  "latitude": 37.5514492,
  "longitude": 126.9209447,
  "reviewCount": 795,
  "placeReviewCount": 1432,
  "hasNaverBooking": false,
  "placeUrl": "https://m.place.naver.com/place/1789177874/home",
  "sourceUrl": "https://map.naver.com/p/search/서울%20카페",
  "query": "서울 카페",
  "scrapedAt": "2026-10-17T20:04:53.074Z"
}

The homepage query string is shortened in this illustration. Actual output preserves the source homepage.

How much does it cost to extract Naver local businesses?

Pricing is pay per event: one start fee per valid run plus one item event per business saved. No separate event is charged for contact fields or review counts. Pricing is $0.005 per start and $0.0014 per business at BRONZE. At these prices 1, 20 and 100 businesses cost $0.0064, $0.033 and $0.145. Other subscription tiers have different item rates.

A valid run that encounters a source challenge may still incur the start fee. Invalid inputs are rejected before charging. Source or billing limits can produce fewer rows than requested. Check the Pricing tab before running.

Limits and reliability

Korean residential proxy and Chromium sessions are built in; no Naver login or private cookies are needed. Searches run sequentially. A stalled initial search has a 15-second navigation/response window, then receives at most one retry with a fresh browser and proxy identity and a 90-second window for slower searches. The windows include response-body consumption; browser startup and cleanup add time outside them. Images, media and fonts are blocked to reduce transfer.

Challenges, unrecognized responses and malformed businesses cause failure rather than misleading empty success. A recognized empty business list yields no rows. A later failed search can leave partial data from earlier searches: inspect run status before treating a dataset as complete.

This Actor does not extract review text, personal-contact enrichment, menus, photos, opening hours, detail pages, booking inventory, exhaustive pages, historical snapshots or built-in change alerts. No CAPTCHA solving or signed-in access is supported.

Integrations

Schedule recurring runs with Apify schedules. Join exports by placeId in your database or spreadsheet to compare snapshots externally. Use webhooks to send successful-run notifications to Make, Zapier or your ETL pipeline. These integrations do not add built-in history or alerts to the Actor.

The run's SUMMARY key-value record reports saved businesses, completed searches and session attempts. Its observed byte counter covers only response Content-Length headers and is not a billing-transfer measure.

API usage

Use a token with access to the Actor. Never publish your token.

curl -X POST "https://api.apify.com/v2/acts/automation-lab~naver-map-business-scraper/runs" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"searchUrls":["https://map.naver.com/p/search/서울%20카페"],"maxItems":20}'

JavaScript using apify-client:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('automation-lab/naver-map-business-scraper').call({
  searchUrls: ['https://map.naver.com/p/search/서울%20카페'], maxItems: 20
});
const { items } = await client.dataset(run.defaultDatasetId).listItems({ limit: 20 });
console.log(items);

Python using apify-client:

import os
from apify_client import ApifyClient
client = ApifyClient(os.environ['APIFY_TOKEN'])
run = client.actor('automation-lab/naver-map-business-scraper').call(
    run_input={'searchUrls': ['https://map.naver.com/p/search/서울%20카페'], 'maxItems': 20})
items = client.dataset(run['defaultDatasetId']).list_items(limit=20).items
print(items)

MCP usage

claude mcp add --transport http apify \
  "https://mcp.apify.com?tools=automation-lab/naver-map-business-scraper"

Claude Desktop, Cursor, and VS Code setup (use each client's HTTP MCP configuration):

{"mcpServers":{"apify":{"url":"https://mcp.apify.com?tools=automation-lab/naver-map-business-scraper"}}}

Authenticate as required by your client. Discover actual tool names and argument schemas with tools/list; Actor selection also exposes run/data helper tools, not exactly one tool. Discovery does not authorize execution.

Example prompt: “Snapshot up to 20 businesses from this Korean Naver Map search URL, then summarize names and addresses.” For an authorized invocation, start once and retain the returned run ID and storage IDs. These IDs are metadata, not business results. If nonterminal, follow the same run using get-actor-run with bounded 2/4/8-second backoff, capped at 10 seconds and a 120-second total deadline. Configure transport timeout above the selected server wait. On timeout recover the known run instead of starting another; at deadline report pending and stop.

After success, fetch the returned dataset with get-dataset-items: pages of 20 source rows, explicit offset and needed fields, at most 100 rows or 64 KiB of serialized UTF-8 content admitted to model context. Enforce bytes host-side, including oversized rows, before model injection; if your client cannot intercept responses, do not claim a hard byte guarantee. Track source offsets, disclose partial/truncated results and leave full exports in storage. Report terminal failures honestly and treat existing rows as partial.

Legality and responsible use

Use public business data responsibly and comply with applicable laws, source terms and privacy obligations. Public availability does not authorize spam or unrestricted reuse. This Actor does not provide legal advice and is not affiliated with Naver.

No AI model processes your inputs or business records. Apify runs and stores the results; its KR residential proxy routes public Naver requests. Company-paid compute, proxy and storage costs are covered by the stated event prices, with no external service account required.

Browser cookies and sessions are ephemeral and discarded when the browser closes. Datasets, input, summary and logs use your Apify storage retention (31 days on our current plan; your workspace policies may differ). Delete run storage through Apify when no longer needed. Logs contain counters and bounded failure reasons, not raw source responses or proxy credentials. For support use the Actor's Issues tab; do not post secrets.

Failed operations send sanitized diagnostic input, exceptions and Actor/build/run IDs to our private GlitchTip service for repair. Secret fields and URL query strings are removed; reports are retained for 30 days. Do not include private information in search URLs.

FAQ

Why fewer rows than maxItems? Only the first business-result page is collected. Sparse searches, duplicate businesses or billing limits can reduce the count.

Can I use English or direct Place URLs? No. This contract requires exact Korean Map search URLs, not Place detail IDs or English queries.

Why did a search fail? A challenge, unsupported search shape or temporary session failure can exhaust both attempts. Check the log and retry later; do not interpret a failed run as an empty market.

Does it enforce city boundaries? No. Naver controls the search results. Use exported addresses or coordinates for your own geographic validation.

Related Actor

For review text from known Place IDs, use Naver Reviews Scraper. That is a separate downstream workflow, not included here.