Naver Place Review Scraper

Reviews of Korean shops, restaurants, clinics and sights live on Naver Map, not Google. This actor reads them for you: give it a few Naver Map searches or place links, and it returns each place's visitor reviews, the Naver Blog and Cafe posts written about it, and a one-row rating summary, as a table you can export to JSON, CSV or Excel.

What it collects

• Visitor reviews: the text, the star score, when it was written and when the visit was, how many times the reviewer has been, the keywords they ticked ("음식이 맛있어요"), and the owner's reply with its date.
• Reviewer context: nickname, profile link, how many reviews they have written and their average score, so you can tell regulars from one-off accounts.
• Blog and Cafe posts: title, text excerpt, link, author and date for the posts Naver attaches to the place.
• A rating summary per place: average score, how many reviews gave each score, and the keywords visitors voted for most.
• Photos when you want them: every review row has a photo count; switch photos on to also get the image links.

Two ways to pick places

Give it	Example	What happens
A Naver Map search	강남 맛집, 홍대 카페, Seoul BBQ	The top places for that search are taken (5 by default)
A place link	https://map.naver.com/p/entry/place/11571707	That place is scraped
A place ID	11571707	Same as the link

Links from map.naver.com, pcmap.place.naver.com, m.place.naver.com and place.naver.com all work. naver.me short links do not, because the ID only appears after a redirect: open the short link in a browser and paste the full address instead. Searches and links can be combined in one run; a place found twice is scraped once.

Naver Shopping product reviews and Naver Blog search results are separate services and are not covered here.

Who uses it

Restaurant and clinic groups watching what visitors say at each branch. Agencies doing Korean local SEO, who need the voted keywords. Tourism teams tracking sentiment at landmarks. Researchers and NLP teams who need Korean review text with star labels.

How a run goes

Searches go through Naver Map's own search box endpoint and return place IDs. For each place the actor then asks Naver's place service for the name, category, address and totals, reads visitor reviews 50 at a time in the order you chose until it hits your limit or the place runs out, adds the blog posts from the place page, and finally the rating summary. Places are worked on in parallel (5 at a time by default) and each place's rows are saved together when it finishes.

Settings

In the order of the input form:

Field	Default	Meaning
searchKeywords	none	Naver Map searches, one per line
maxPlacesPerKeyword	5	Top results kept per search, 1 to 50
placeUrls	demo place	Place links or numeric IDs
includeReviewStats	on	One rating-summary row per place
includeBlogReviews	on	Blog and Cafe post rows
maxBlogReviewPages	10	Blog posts kept per place, in tens
reviewSort	NEWEST	NEWEST or RANKING (Naver's default order)
maxReviewPages	1	Visitor reviews per place, in fifties
maxReviews	none	Exact visitor review cap; replaces the page count when set
includeReviewPhotos	off	Image links on visitor reviews
maxConcurrency	5	Places scraped at once, 1 to 10
proxy	Apify residential	Naver throttles datacenter IPs

Sample inputs

The 100 newest reviews of one place, nothing else:

{
  "placeUrls": ["11571707"],
  "maxReviews": 100,
  "includeBlogReviews": false,
  "includeReviewStats": false
}

Rating summary and the latest 50 reviews for the top 10 Gangnam restaurants:

{
  "searchKeywords": ["강남 맛집"],
  "maxPlacesPerKeyword": 10,
  "maxReviewPages": 1
}

Two places with photo links and blog posts:

{
  "placeUrls": ["11571707", "https://map.naver.com/p/entry/place/38345004"],
  "maxReviewPages": 4,
  "includeReviewPhotos": true,
  "maxBlogReviewPages": 3
}

What comes back

Every row has a reviewType:

• visitor: one visitor review
• blog: one Naver Blog or Cafe post
• place_stats: the rating summary of one place

Each row also repeats the place it belongs to (ID, name, category, address, average score, total review count and link), so any single row can be read on its own and rows from many places can share one sheet.

Real rows from a run on Gyeongbokgung Palace, shortened:

{
  "placeName": "경복궁",
  "reviewType": "visitor",
  "reviewRating": 5,
  "reviewDate": "10.21.수",
  "reviewText": "한한령 중국 한국 여행 금지 풀렸나 봐요...",
  "votedKeywords": ["사진이 잘 나와요", "볼거리가 많아요", "방문객이 많아요", "규모가 커요"],
  "visitDate": "10.21.수",
  "visitCount": 2,
  "reviewPhotoCount": 14,
  "reviewId": "6ad874ae656738da5f4c1e6c",
  "reviewerName": "백수인",
  "placeCategory": "궁궐",
  "placeAddress": "서울 종로구 사직로 161 경복궁",
  "placeRating": 4.69,
  "placeTotalReviews": 12539,
  "placeId": "11571707",
  "placeUrl": "https://pcmap.place.naver.com/place/11571707"
}

{
  "placeName": "경복궁",
  "reviewType": "blog",
  "blogTitle": "아이와 서촌산책코스 | 경복궁 은행나무...",
  "blogUrl": "https://m.blog.naver.com/ssongji337/224448694403",
  "blogType": "블로그",
  "bloggerName": "랑새",
  "blogDate": "2일 전",
  "placeId": "11571707"
}

{
  "placeName": "YTN서울타워",
  "reviewType": "place_stats",
  "averageRating": 4.47,
  "starDistribution": [
    { "score": null, "count": 1 },
    { "score": 2.5, "count": 2 },
    { "score": 3, "count": 12 },
    { "score": 3.5, "count": 10 },
    { "score": 4, "count": 27 },
    { "score": 4.5, "count": 16 },
    { "score": 5, "count": 88 }
  ],
  "placeId": "38345004"
}

Dates are kept the way Naver prints them (10.21.수 is 21 October, a Wednesday; 2일 전 is two days ago).

Field groups

• The place: placeId, placeName, placeCategory, placeAddress, placeRating, placeTotalReviews, placeUrl
• The visitor review: reviewId, reviewText, reviewRating, reviewDate, visitDate, visitCount, votedKeywords, visitCategories, reviewTags, reviewPhotoCount, reviewPhotos
• The reviewer: reviewerName, reviewerProfileUrl, reviewerImageUrl, reviewerReviewCount, reviewerAvgRating
• The owner's answer: reviewReplyText, reviewReplyDate
• The blog post: blogTitle, blogContents, blogUrl, blogType, bloggerName, blogDate, blogThumbnailUrl
• The summary: averageRating, starDistribution, topVotedKeywords
• The run: scrapedAt

Fields that do not apply to a row type are null (a blog row has no star score).

starDistribution counts the rated reviews per half-star score (0.5 to 5), lowest first; reviews left without stars are not in it. Naver sends these counts without their scores and drops the scores nobody gave, so the actor works the scores out from Naver's own average. A bucket it cannot place for certain keeps score: null; that happens only to small buckets at the low end, like the single review above.

Billing and the free plan

One review-scraped event per visitor row and per blog row. Rating-summary rows are free. If the run reaches your spending limit it stops reading new reviews and saves what it has.

On Apify's free plan a run keeps up to 10 visitor reviews and one page of blog posts per place, and takes 2 places per search.

Questions

Do I need a proxy? A place or two works without one. Past that, Naver starts refusing requests from a single IP, which is why the default is Apify's residential proxy.

Pages or an exact number? maxReviewPages counts in fifties and is the simple choice. Set maxReviews when you need an exact number such as 120; it then replaces the page count.

Why fewer visitor reviews than the place's review count? The actor reads every review in Naver's review list, up to your limit. That list can be shorter than the counter on the place page: for N Seoul Tower the counter said 510 and the list held 416 on 24 Oct 2026. Naver does not list the rest.

Why fewer blog posts than Naver's total? The actor takes the posts Naver shows on the place page. For places with thousands of posts that is a limited batch, not the full archive.

Can I get English reviews? Reviews come in the language the visitor wrote, almost always Korean. Voted keywords are fixed Korean phrases Naver offers to every reviewer.

Why did a link get skipped? It was a naver.me short link or a page that is not a place (a search result list, a blog). The run log names each skipped line.

Support

• Issues tab of this actor
• Website: muhamed-didovic.github.io
• Email: muhamed.didovic@gmail.com

Additional Services

Need a recurring export, a filter the actor does not have, or a one-off dataset of Naver Place reviews? Write to muhamed.didovic@gmail.com.

Explore More Scrapers

More actors for real estate, jobs, B2B data, social media and e-commerce: muhamed-didovic.github.io.

🤖 For AI Agents & LLM Apps

memo23/naver-place-review-scraper returns Naver Map (네이버 플레이스) visitor reviews, blog posts and rating summaries for Korean businesses.

Minimal input:

{ "placeUrls": ["11571707"], "maxReviews": 20 }

• Input: placeUrls (links or numeric IDs) and/or searchKeywords (Naver Map searches).
• Rows: reviewType is visitor, blog or place_stats; every row carries placeId and placeName.
• Main visitor fields: reviewText, reviewRating, reviewDate, visitDate, votedKeywords, reviewReplyText.
• Billing: one event per visitor or blog row; summary rows free. Free plan: 10 reviews per place.

⚠️ Disclaimer

This actor is independent and is not made, endorsed or sponsored by NAVER Corporation. Naver and Naver Place are trademarks of NAVER Corporation. The actor reads publicly visible review data. Reviewer nicknames and profile links are personal data under Korea's Personal Information Protection Act (PIPA) and the GDPR; you are responsible for having a legal basis to collect and use them, and for following Naver's terms.

SEO Keywords

naver place scraper, naver place review scraper, naver map reviews, 네이버 플레이스 리뷰, 네이버 지도 리뷰 크롤링, korean review scraper, naver visitor reviews export, naver blog reviews, naver cafe posts, korean local business reviews, naver rating summary, naver voted keywords, korean restaurant reviews, gangnam restaurant reviews, korean local seo, naver review sentiment, korean review dataset, naver map data export