Tokopedia Shop Reviews Scraper

Give it one or more Tokopedia shop URLs and it crawls every product of each shop and collects all of their reviews — text, star rating, reviewer, date, photos, videos, and seller replies — as clean, structured JSON ready for spreadsheets, BI tools, or sentiment analysis.

Perfect for brand monitoring, competitor research, product feedback analysis, and review datasets.

Features

• 🏬 Whole-shop crawl — point it at a shop, get reviews for every product.
• ⭐ Full review detail — rating, text, reviewer, timestamp, likes, variant, images, and videos.
• 💬 Seller replies — optionally captured as separate rows to measure brand responsiveness.
• 📊 Per-product summaries — average rating and 1–5 star distribution per product.
• 🗓️ Date & rating filters — only pull reviews since a date, or within a star range.
• 🔁 Incremental mode — remember what you've already scraped and only fetch new reviews on the next run (ideal for scheduled monitoring).
• 🧹 Clean, stable schema — every field documented; drop straight into your pipeline.

Input

Field	Type	Default	Description
shopUrls	string[]	—	Tokopedia shop URLs (https://www.tokopedia.com/<slug>).
shopIds	string[]	—	Numeric shop ids — use if URL resolution fails.
maxProductsPerShop	int	0	Cap products per shop. 0 = all.
maxReviewsPerProduct	int	0	Cap reviews per product. 0 = all.
onlyProductsWithReviews	bool	false	Skip products the listing marks as 0 reviews (listing count is unreliable, so off by default).
reviewSort	enum	Most helpful	Most helpful / Newest / Oldest / Highest / Lowest.
reviewsSince	date	""	Keep only reviews on/after this date. Sorts newest-first and stops a product once older reviews are reached.
includeProductSummaries	bool	true	Emit one aggregate row per product to the product-summaries dataset.
includeSellerReplies	bool	true	Emit seller replies as extra rows (source_label: Tokopedia Seller Reply).
dedup	bool	false	Remember review ids across runs and skip seen ones. For scheduled monitoring.
minRating / maxRating	int	0	Keep only reviews within this 1–5 range. 0 = no filter.
productConcurrency	int	3	Products scraped in parallel.
requestDelayMs	int	400	Polite delay between requests.
proxyConfiguration	proxy	off	Apify Proxy settings.

At least one of shopUrls / shopIds is required. When reviewsSince or dedup is enabled, review sort is forced to newest-first so early-stop can work.

Example input

{
  "shopUrls": ["https://www.tokopedia.com/wahanakaryastore"],
  "maxReviewsPerProduct": 200,
  "reviewsSince": "2026-01-01",
  "includeSellerReplies": true
}

Output

One row per review (and per seller reply when enabled) in the default dataset:

{
  "external_id":  "tokopedia-review-1492210321",
  "author_handle":"L***i",
  "author_name":  "L***i",
  "body":         "Trimakasih paket nya udah sampai...",
  "posted_at":    "2025-05-29T17:07:08.000Z",
  "url":          "https://www.tokopedia.com/wahanakaryastore/solder-...",
  "source_label": "Tokopedia Review",
  "likes":        0,          // helpful/like count
  "comments":     0,          // 1 if the seller replied
  "reach":        0,          // reserved (always 0)
  "rating":       5,          // 1-5
  "raw": {
    "platform": "tokopedia",
    "feedbackId": "1492210321",
    "productId": "100230143849",
    "productName": "...",
    "variantName": "Solder 60w+Timah 5m",
    "shopId": "7494083584907249524",
    "shopName": "WahanaKaryaStore",
    "isAnonymous": true,
    "relativeTime": "Lebih dari 1 tahun lalu",
    "sellerReply": null,
    "images": ["https://..."],
    "videos": [],
    "badRatingReason": null
  }
}

Every field is documented in the actor's output schema, so the Output tab shows a clean, labelled table you can export to JSON, CSV, Excel, or via API.

Extra datasets

• product-summaries (when includeProductSummaries) — one row per product: totalReviewsReported, reviewsCollected, averageRatingCollected, ratingDistribution (1–5 star counts), partial flag. The star distribution is computed from the reviews actually fetched, so it's exact only on a full pull (no maxReviewsPerProduct / reviewsSince / dedup).
• tokopedia-reviews-state key-value store (when dedup) — persists seen review ids across runs.

Notes

Tokopedia sits behind Akamai Bot Manager. Plain datacenter IPs may get challenged. If you see repeated failures, enable Apify Proxy (RESIDENTIAL, country ID) in the input.

The scraper reads Tokopedia's public GraphQL endpoints (no login required). If Tokopedia changes its API and a run starts erroring, please report it — signatures occasionally change and the queries need re-capturing.

Run locally

npm install
# create storage/key_value_stores/default/INPUT.json with your input, then:
npm start