Yelp Reviews API | Extract Ratings, Review Text & Reviewer Data (MCP-ready)

Scrape Yelp reviews by place ID and get clean JSON back. Star rating, full review text, date, reviewer name, location and stats, photos, owner replies, and helpful votes. Sort by date or rating, filter by stars, search within reviews, and paginate. MCP-ready for Claude, ChatGPT, Cursor, and other AI agents.

Give the Actor a Yelp place ID and it returns that business's reviews as structured JSON: the star rating, full review text, the date, the reviewer's name, location and activity stats, attached photos, the business owner's replies, and useful/funny/cool vote counts. Sort newest-first or by rating, keep only certain star ratings, search reviews for a keyword, or pull Yelp's not-recommended (filtered) reviews. Pages return up to 49 reviews each.

This is the reviews stage of a 3-actor Yelp suite. Get place IDs from the Yelp Search API, fetch full profiles with the Yelp Business Details API, and pull reviews here. Predictable per-page pricing, structured JSON, no browsers or captchas.

---

What this Actor returns

Each page is one dataset item containing a reviews array. Each review includes:

• Rating & text - star rating, the full review comment, and its detected language.
• Date - when the review was posted.
• Reviewer - name, location, and activity stats (friends, photos, review count).
• Photos - any images attached to the review.
• Owner replies - the business's responses to the review.
• Feedback - useful / funny / cool vote counts, plus tags.

---

Use with Claude, ChatGPT, Cursor & other AI agents (MCP)

This Actor is a first-class tool on the Apify MCP Server. Any MCP-compatible AI agent - Claude (Desktop, Web, Code), ChatGPT, Cursor, VS Code, Cline, Windsurf, Kilo Code, Opencode, Glama - can discover and call it in natural language.

What an AI agent does with this:

User: "Summarize the 1-star reviews for this restaurant on Yelp."

Agent resolves the business to a place ID via the Yelp Search API, calls this Actor with {"place_id": "ED7A7vDdg8yLNKJTSVHHmg", "rating": "1", "sort_by": "date_desc", "max_pages": 2}, and summarizes the complaints.

New to Claude? Claude Code and the Claude desktop app (which runs Cowork) both come with a free trial: https://claude.ai/referral/uIlpa7nPLg

Quick setup - Claude Desktop

Add this to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "apify": {
      "command": "npx",
      "args": ["-y", "@apify/actors-mcp-server"],
      "env": {
        "APIFY_TOKEN": "YOUR_APIFY_API_TOKEN"
      }
    }
  }
}

Restart Claude Desktop, then ask something like "What do recent Yelp reviewers say about the service here?"

Quick setup - Cursor / VS Code / Cline / Windsurf

These editors support dynamic tool discovery, so after the first call this Actor is registered as a named tool for the rest of the session. Point your MCP client at:

https://mcp.apify.com

…with header Authorization: Bearer YOUR_APIFY_API_TOKEN. Full setup: Apify MCP integration docs.

Quick setup - ChatGPT (and other static MCP clients)

ChatGPT, Gemini CLI, and Amazon Q connect through the same https://mcp.apify.com endpoint and call this Actor via the generic call-actor tool.

---

Use cases

• Reputation monitoring - track new reviews and ratings for your business or competitors.
• Sentiment analysis - feed review text into an LLM to surface themes and complaints.
• Service quality tracking - filter to 1-2 star reviews to find recurring issues.
• Owner-response audits - see which reviews got a reply and how fast.

---

Input parameters

Parameter	Type	Required	Default	Description
place_id	string	Yes	-	Yelp encoded place ID (e.g. ED7A7vDdg8yLNKJTSVHHmg) - the first entry of the place_ids array from the Yelp Search API.
sort_by	string (enum)	No	relevance	relevance_desc, date_desc, date_asc, rating_desc, rating_asc, elites_desc.
rating	string	No	-	Filter by stars: 5 or a list like 4,5.
q	string	No	-	Keep only reviews mentioning this keyword.
hl	string	No	en	Two-letter language code.
not_recommended	boolean	No	false	Return Yelp's not-recommended (filtered) reviews instead.
max_pages	integer	No	1	Pages to fetch (~49 reviews each). 0 = unlimited (cap 20).

---

Example output (one item per page)

{
  "page_number": 1,
  "result_timestamp": "2026-05-26T10:30:00.123456",
  "search_parameters": { "place_id": "ED7A7vDdg8yLNKJTSVHHmg", "sortby": "date_desc", "max_pages": 1 },
  "search_metadata": { "pages_processed": 1, "max_pages_set": 1, "pagination_limit_reached": false, "total_reviews_estimate": 847 },
  "reviews": [
    {
      "position": 1,
      "rating": 5,
      "date": "2026-05-20",
      "user": { "name": "Jane D.", "location": "New York, NY", "stats": { "friends": 120, "photos": 45, "reviews": 88 } },
      "comment": { "text": "The almond croissant was incredible...", "language": "en" },
      "photos": [],
      "feedback": { "useful": 3, "funny": 0, "cool": 1 },
      "owner_replies": [],
      "tags": []
    }
  ]
}

---

Pricing

This Actor uses transparent pay-per-event pricing:

Event	Price	When
Setup	$0.02	Once per run
Page processed	$0.02	Per page of reviews fetched (~49 reviews)

Fetching 1 page (~49 reviews) costs about $0.04; 5 pages (~245 reviews) costs about $0.12.

---

How to get started

1. Open the Actor on the Apify Store.
2. Enter a place_id, set sort/filter options and max_pages, and click Start.
3. Read results from the dataset (JSON, CSV, Excel) or via the Apify API.

---

Code example (Python + MCP)

Want a runnable quick-start? The public example repo has a Python (uv) script plus MCP install guides for Claude (Desktop, Code, Web) and Cursor:

github.com/johnisanerd/Apify-Yelp-API

It shows how the Yelp Search, Business Details, and Reviews APIs chain together - a search returns the place_ids that feed the other two.

FAQ / Troubleshooting

• Where does the place ID come from? From the Yelp Search API - every listing includes a place_ids array. Use the encoded ID (the first entry, e.g. ED7A7vDdg8yLNKJTSVHHmg), not the human-readable alias, for the reviews endpoint.
• How many reviews per page? Up to 49. Increase max_pages to fetch more.
• Can I get only 5-star (or 1-star) reviews? Yes - set rating to 5 or 1, or a list like 4,5.
• What are not-recommended reviews? Yelp hides some reviews behind a "not recommended" filter. Set not_recommended: true to fetch those instead.
• Is this reliable? Yes - it calls a structured data API, not a headless browser, so there are no captchas or layout breakages.

Learn more about the Apify MCP integration.

Last Updated: 2026.05.29