OpenTable Reviews & Restaurants Scraper, No Login

Paste OpenTable restaurant links and get their guest reviews with overall, food, service, ambience and value ratings, the dined date and the restaurant's reply, or search OpenTable restaurants by city and get cuisine, price band, rating, review count, address and phone. No login, cookies or API key.

Use it to track what diners say about your restaurant or your competitors, or to build a restaurant list for a city. The smallest run is one restaurant link with 20 reviews: it takes about 20 seconds and costs about $0.041.

Table of contents

• What it does
• How it works
• Quick start
• Use cases
• Input
• Output
• Pricing
• Limits
• API, export and integrations
• Troubleshooting
• FAQ
• Support

What it does

The Actor reads public pages of OpenTable, the restaurant booking site. It has two modes:

• Reviews — you add restaurant links or IDs. For each restaurant you get one restaurant row with its details, then its reviews, newest first (or highest / lowest rated first).
• Search — you add a city or area. You get the restaurants OpenTable lists there, one row per restaurant, with details from each restaurant page.

All rows go to one Dataset with two views: Reviews and Restaurants. A view chooses the columns; it does not filter rows, so filter by type (review or restaurant) when you export. A run summary goes to the Key-value store as RUN_SUMMARY.

This Actor is not made by OpenTable and is not connected to OpenTable. It reads only what OpenTable shows to every visitor without login.

Good fit

• Restaurant owners and marketing teams who watch their own and competitors' OpenTable reviews, ratings and replies.
• Analysts who compare restaurants in a city by cuisine, price band, rating and review count.
• Scheduled monitoring: with reviewsSince a run reads only new reviews and stops at the first older one.

How it works

1. The Actor reads OpenTable's public pages and its reviews feed over plain HTTPS, the same data any visitor sees without login. No browser starts while OpenTable lets these requests in, so runs are fast.
2. In Reviews mode it opens each restaurant page, saves the restaurant row, then reads its reviews page by page until your limits or your reviewsSince date. In Search mode it reads OpenTable's restaurant list for your location and, if details are on, opens each restaurant page.
3. If OpenTable refuses these requests, at the start or later in the run, the Actor tries a fresh session, then US residential addresses, and then a real Chrome browser for the rest of the run. Plain HTTP gets up to 4 residential sessions; Chrome then gets one more residential address and 30 seconds for its first answer, so a run refused everywhere stops in about a minute and a half. This is included in the price. A server error at OpenTable (HTTP 5xx) is retried 3 times and then stops the run; it does not switch routes.
4. It saves review and restaurant rows to the Dataset, with status on each restaurant row, and a run summary as RUN_SUMMARY. Its route field says how OpenTable was read: http (no browser) or browser (Chrome was needed).

Quick start

1. Open the Input tab.
2. Keep What to scrape (mode) as Reviews of restaurant URLs.
3. In Restaurant URLs or IDs (restaurantUrls) keep the example link or paste your own, e.g. https://www.opentable.com/r/brooklyn-chop-house-new-york.
4. Keep Max reviews per restaurant (maxReviewsPerRestaurant) at the prefilled 20.
5. Click Start.
6. Open the Output tab, view Reviews, and check rating, text and dinedAt.

Expected result: one restaurant row and 20 review rows, in about 20 seconds.

Use cases

Buyer job	Input path	Output entity	Key fields	Next integration
Read guest feedback for a restaurant	restaurantUrls, maxReviewsPerRestaurant	review	rating, ratingFood, ratingService, text, dinedAt	Export CSV or Excel
Check how a restaurant answers reviews	restaurantUrls, sortReviewsBy: "lowest"	review	rating, restaurantReply, restaurantReplyAt	Spreadsheet or BI tool
Get new reviews every week	restaurantUrls, reviewsSince: "7 days"	review	reviewId, submittedAt, rating, text	Apify schedule + API
Build a restaurant list for a city	mode: "search", location	restaurant	name, cuisine, priceBand, rating, reviewCount, phone	CRM or map

Input

Input fields

Field	Plain name	Type	Required	Default / prefill	Allowed values	Effect	Example	Recommendation
mode	What to scrape	string	no	default reviews, prefill reviews	reviews, search	Reviews of the links below, or restaurants for a location	"search"	If you fill only location, the Actor switches to Search; if only links, to Reviews
restaurantUrls	Restaurant URLs or IDs	array of strings	in Reviews mode	prefill: 1 New York link	up to 200; /r/<name> links, /restaurant/profile/<id> links, links with ?rid=, numeric IDs, any OpenTable country site	Which restaurants to read	["https://www.opentable.com/r/brooklyn-chop-house-new-york", "1017331"]	One run reads the first 40; the rest are listed as not_started (free). Duplicates are read once
location	Location	string	in Search mode	none	up to 200 characters	City, neighborhood or area to search	"Chicago"	Use the name you would type on OpenTable
maxReviewsPerRestaurant	Max reviews per restaurant	integer	no	default 100, prefill 20	1–10,000	Most reviews for each restaurant	500	Each returned review is one paid event
maxItems	Max results	integer	no	empty: 1,000 reviews (Reviews) or 30 restaurants (Search)	Reviews 1–50,000; Search 1–1,000	Most rows in the whole run	30	In Search with details, keep it at 38 or less for details on every row
sortReviewsBy	Review order	string	no	newest	newest, highest, lowest	Which reviews come first	"lowest"	Use lowest to find complaints
reviewsSince	Only reviews since	string	no	none	a date like 2026-01-31, an ISO time, or 30 days, 2 weeks, 6 months, 1 year	Only reviews submitted on or after this time	"30 days"	With newest order the run stops at the first older review, so you pay only for new ones
includeRestaurantDetails	Include restaurant details	boolean	no	true	true, false	Search only: open each restaurant page for address, phone, hours and sub-ratings	false	Turn off for a faster list with name, cuisine, price band, rating and review count

Empty input: the Console fills in the prefill. Through the API, an input with no links and no location fails with a clear invalid_input message; only the start is charged.

Interacting settings

When	Input A	Input B	Effective behavior	Example	Cost/result effect
Many restaurants	maxReviewsPerRestaurant	maxItems	The run stops at whichever limit comes first	10 links × 500, maxItems 1,000	At most 1,000 review events
Monitoring	reviewsSince	sortReviewsBy: "newest"	Reading a restaurant stops at the first older review	"7 days"	No new reviews → only the start and restaurant rows
reviewsSince with another order	reviewsSince	sortReviewsBy: "highest"	Older reviews are skipped but still read	"30 days" + highest	Slower; you still pay only for returned reviews
Search with details	includeRestaurantDetails: true	maxItems above 38	Rows past the page limit come with search-list data, status: "partial" and a warning	maxItems 60	Same restaurant price per row
Both filled	restaurantUrls	location	mode decides which one is used	mode: "search"	The other field is ignored

Input examples

Reviews of one restaurant (prefill)

{
  "mode": "reviews",
  "restaurantUrls": ["https://www.opentable.com/r/brooklyn-chop-house-new-york"],
  "maxReviewsPerRestaurant": 20
}

Expected behavior: the Actor opens the restaurant page and reads its 20 newest reviews. Expected output: 1 restaurant row and 20 review rows with rating, text, dinedAt, restaurantReply. Limit or cost note: about 20 seconds and about $0.041 (start $0.022 + restaurant $0.003 + 20 × $0.0008).

Restaurants in a city

{
  "mode": "search",
  "location": "Chicago",
  "maxItems": 30
}

Expected behavior: the Actor reads OpenTable's restaurant list for Chicago and opens each restaurant page. Expected output: 30 restaurant rows with name, cuisine, priceBand, rating, phone. Limit or cost note: about 40 seconds and about $0.11 (start $0.022 + 30 × $0.003).

Output

Output fields

Review rows (type: "review"):

Field	Meaning	Type	Can be empty?	Example / enum	Source / notes
type	Row type	string	no	review	—
restaurantId, restaurantName, restaurantUrl	The restaurant this review is about	integer / string	no	1017331	—
reviewId	OpenTable review ID	string	rarely	OT-1017331-261886-120047159362	Use it to remove duplicates between runs
rating	Overall rating 1–5	number	yes, if the diner gave none	5	—
ratingFood, ratingService, ratingAmbience, ratingValue	Sub-ratings 1–5	number	yes	4	null = not rated, not 0
noise	The diner's noise answer	string	yes	QUIET, MODERATE, ENERGETIC, DO_NOT_RECALL	—
text	Review text	string	yes	"loved the ambiance…"	HTML entities decoded, line breaks kept
dinedAt	When the diner ate there	string	yes	2026-09-30T18:30:00-04:00	Restaurant local time with offset
submittedAt	When the review was submitted	string	yes	2026-10-01T21:53:34Z	UTC
reviewerName, reviewerMetro, reviewerReviewCount, reviewerIsVip	Public reviewer profile	string / integer / boolean	yes	Diner1, Chicago, 8, false	Public display name only
restaurantReply, restaurantReplyAt	The restaurant's public reply	string	yes, when there is no reply	—	—
helpfulUp, helpfulDown	Helpful votes	integer	yes	0	—
verified, fromPreviousConcept	Verified diner; written for an earlier concept	boolean	yes	true, false	—
photoCount	Number of photos	integer	no	0	Photos are not downloaded
source, fetchedAt	Always opentable; read time (UTC)	string	no	2026-10-14T15:29:52.739Z	—

Restaurant rows (type: "restaurant"):

Field	Meaning	Type	Can be empty?	Example / enum	Source / notes
status	Row result	string	no	complete, partial, not_found, error	See Output entities
restaurantId, name, url	Restaurant identity	integer / string	null only on not_found / error	Brooklyn Chop House - Downtown	—
cuisine, cuisines	Main cuisine; all cuisines	string / array	yes	Steakhouse	—
priceBand, priceLevel, currencySymbol	Price per person	string / integer / string	yes	$31 to $50, 3, $	—
rating, ratingFood, ratingService, ratingAmbience, ratingValue	Average ratings 1–5	number	yes	4.1	Based on recent reviews, as on the OpenTable page
ratingNoise, ratingCount, ratingDistribution	Noise 1–3; recent reviews counted; stars 1–5	number / integer / object	yes	{"1": 9, …, "5": 74}	Details only
reviewCount	All-time written reviews	integer	yes	3055	—
address, street, city, state, postalCode, country, countryCode	Address	string	yes	150 Nassau St, New York, NY 10038, United States	—
neighborhood, metro, latitude, longitude	Area and position	string / number	yes	TriBeCa, 40.71	—
phone, website	Contacts	string	yes	(212) 619-1200	Details only
hours, diningStyle, dressCode, executiveChef, paymentOptions	Restaurant facts	string / array	yes	Fine Dining	Details only
description, reviewSummary, awards, imageUrl	Text, OpenTable review summary, awards, photo	string / array	yes	—	Details only
permanentlyClosed	OpenTable marks it as closed	boolean	yes	false	—
detailsIncluded	The restaurant page was read	boolean	no	true	false = search-list data only
searchLocation, searchPosition	Search mode: location and position	string / integer	yes	Chicago, 1	null in Reviews mode
input, error	Your input entry; why the row is not complete	string	yes	Not found on OpenTable	Set on partial, not_found, error

null means OpenTable did not give the value. It never means 0.

Output entities and surfaces

Situation	Dataset/API/KVS/live view	Result shape	User action
Normal item	Dataset, views Reviews and Restaurants	One restaurant row, then its review rows; or one restaurant row per search result	Open, export or call the Dataset API
No reviews or no search results	Dataset + RUN_SUMMARY	Restaurant row without reviews, or no rows; run SUCCEEDED with a warning	Check the link, reviewsSince or location
Restaurant not on OpenTable	Dataset	restaurant row with status: "not_found", input, error	Check the link
Link that is not OpenTable	Dataset	restaurant row with status: "error" (free)	Fix the link
Details skipped or restaurant page failed (Search)	Dataset	status: "partial", detailsIncluded: false, error	Run again with a smaller maxItems
More than 40 restaurants (Reviews)	RUN_SUMMARY	Run SUCCEEDED with status: "partial"; extra restaurants not_started (free), warning	Start another run with the rest
OpenTable blocked, did not answer or kept answering HTTP 5xx	RUN_SUMMARY, run status	Run FAILED with stopReason: "SOURCE_ERROR"; rows saved before the block stay	Try again later
Max charge, timeout or abort	RUN_SUMMARY	Run SUCCEEDED, status: "partial", saved rows stay	Raise the limit and run again

Output example

{
  "type": "review",
  "restaurantId": 1017331,
  "restaurantName": "Brooklyn Chop House - Downtown",
  "restaurantUrl": "https://www.opentable.com/r/brooklyn-chop-house-new-york",
  "reviewId": "OT-1017331-261886-120047159362",
  "rating": 5,
  "ratingFood": 5,
  "ratingService": 5,
  "ratingAmbience": 4,
  "ratingValue": 4,
  "noise": "ENERGETIC",
  "text": "loved the ambiance. Alex, our waiter, was FABULOUS! My filet was cooked to perfection…",
  "dinedAt": "2026-09-30T18:30:00-04:00",
  "submittedAt": "2026-10-01T21:53:34Z",
  "reviewerName": "Diner1",
  "reviewerMetro": "County Durham",
  "reviewerReviewCount": 8,
  "reviewerIsVip": false,
  "helpfulUp": 0,
  "helpfulDown": 0,
  "restaurantReply": null,
  "restaurantReplyAt": null,
  "verified": true,
  "fromPreviousConcept": false,
  "photoCount": 0,
  "source": "opentable",
  "fetchedAt": "2026-10-14T15:29:52.739Z"
}

This is one full review row from the prefill run; the text is shortened and the reviewer name is replaced. restaurantReply: null means the restaurant did not reply. A field is null when OpenTable does not show it for this review.

Use case → fields

Buyer job	Input path	Output entity	Key fields	Next integration
Watch guest feedback	restaurantUrls	review	rating, ratingFood, ratingService, text, dinedAt	CSV or Excel export
Track replies to bad reviews	restaurantUrls, sortReviewsBy: "lowest"	review	rating, restaurantReply, restaurantReplyAt	Spreadsheet or BI tool
Collect only new reviews	restaurantUrls, reviewsSince	review	reviewId, submittedAt, text	Apify schedule + Dataset API
Compare restaurants in a city	mode: "search", location	restaurant	cuisine, priceBand, rating, reviewCount, phone	CRM or map

Pricing

Pricing unit

The Actor uses pay-per-event pricing. You pay for:

Event	When	Price (free plan)	Price (Gold plan and higher)
Actor start	Once per run, 2 events (the Actor always runs with 2 GB)	$0.022 per run	$0.022 per run
Review	Each review saved	$0.0008	$0.00068
Restaurant	Each restaurant row saved (search result, or the restaurant of the reviews you asked for)	$0.003	$0.0026
Restaurant not found	A link or ID that OpenTable does not have (not_found row)	$0.0015	$0.0015

Bronze and Silver plans pay prices between these. The start fee scales with run memory: $0.011 per GB, and the Actor always runs with 2 GB, so $0.022 per run. Platform usage (compute) and the residential address used when OpenTable refuses the Actor's own are included; you do not pay them separately.

Free: links that are not OpenTable (error rows) and restaurants left not_started. A partial restaurant row is charged as a normal restaurant row. A run with no results (no reviews since your date, or a city with no restaurants) costs only the start, $0.022. A failed run charges only the rows it saved before the failure, plus the start.

Cost examples

Scenario	Input scope	Expected item/event count	Base charge	Optional charge	Stop/cap behavior
Prefill	1 link, 20 reviews	1 restaurant + 20 reviews	$0.022 start	$0.003 + 20 × $0.0008 = $0.019	about 20 s
One restaurant, 1,000 reviews	1 link, maxReviewsPerRestaurant 1,000	1 + 1,000	$0.022	$0.003 + $0.80
Weekly check, no new reviews	1 link, reviewsSince: "7 days"	1 restaurant	$0.022	$0.003	stops at the first older review
City list	location, maxItems 30	30 restaurants	$0.022	30 × $0.003 = $0.09	about 40 seconds
City with no results	location OpenTable does not know	0	$0.022	none	SUCCEEDED with a warning

Set Maximum cost per run in the run options to cap your spend: the Actor stops before it goes over, keeps what it saved, and marks the run partial.

Limits

• No reservations, table availability, menus or review photos (only photoCount).
• Only public data. The Actor does not log in and does not return private diner data.
• OpenTable shows at most 10,000 reviews of one restaurant in one order. For more, run again with another sortReviewsBy.
• One run opens at most 40 OpenTable pages, because OpenTable blocks more. In Reviews mode that is 40 restaurants per run (reviews themselves do not count). In Search mode with details, about 38 restaurants get full details; the rest come with search-list data only.
• Search returns at most 1,000 restaurants per run (the maxItems maximum). OpenTable may list fewer unique restaurants for a big city.
• If OpenTable refuses every address the Actor tries in about a minute and a half, the run fails with a clear message; it does not return an empty result as success. Residential use per run is capped (after a plain-HTTP refusal, Chrome gets one residential address for about the first page); past the cap the run goes on from its own address and may stop early with the rows read so far.

API, export and integrations

• API: start a run with POST https://api.apify.com/v2/acts/cliqtomedia~opentable-reviews-scraper/runs and read rows from the run's default Dataset (Apify API docs).
• Export: download the Dataset as JSON, CSV, Excel or XML from the Output tab.
• Scheduling: create an Apify schedule with reviewsSince (for example "7 days") to get only new reviews every week. Use reviewId to remove duplicates.

Use with AI agents (MCP)

• MCP server URL with only this Actor: https://mcp.apify.com?tools=cliqtomedia/opentable-reviews-scraper (Apify MCP docs).

• Smallest useful input:

  {"mode": "reviews", "restaurantUrls": ["https://www.opentable.com/r/brooklyn-chop-house-new-york"], "maxReviewsPerRestaurant": 20}
  

• Key output fields: type (review or restaurant), rating (overall 1–5), text (review text), dinedAt (visit date), status (restaurant row result: complete, partial, not_found, error).

• The prefill run takes about 20 seconds and costs about $0.041. Limits: up to 40 restaurants per run, up to 10,000 reviews per restaurant in one order, maxItems up to 1,000 in Search mode.

Troubleshooting

The run failed with "OpenTable could not be read"

Symptom: the run is FAILED and RUN_SUMMARY has stopReason: "SOURCE_ERROR". Likely reason: OpenTable refused every address the Actor tried in about a minute and a half, or did not answer in time. Check: open RUN_SUMMARY and see how many rows were saved before the failure. Action: start the run again a few minutes later. Rows saved before the failure stay in that run's Dataset.

Some restaurants are not_started

Symptom: the warning says some restaurants were not started. Likely reason: you added more than 40 restaurants. One run opens at most 40 OpenTable pages. Check: RUN_SUMMARY.restaurants lists every input with its status. Action: start another run with the restaurants that are not_started. They are free.

Search rows have no phone or address

Symptom: rows have status: "partial" and detailsIncluded: false. Likely reason: maxItems is above 38 with details on, or one restaurant page could not be read. Check: the row error field says which case it is. Action: run again with maxItems 38 or less, or search a smaller area.

A restaurant has fewer reviews than reviewCount

Symptom: reviewsSaved in RUN_SUMMARY is lower than reviewCount. Likely reason: reviewCount counts all written reviews ever; OpenTable shows at most 10,000 in one order, and reviewsSince or maxItems can stop earlier. Action: raise the limits or run again with another sortReviewsBy.

FAQ

Do I need an OpenTable account, cookies or a proxy?

No. The Actor reads public pages without login, cookies or API key. You do not set a proxy: if OpenTable refuses the Actor's own address, it switches to a US residential address by itself, at no extra charge.

Which OpenTable sites work?

Links from any OpenTable country site, for example opentable.com, opentable.co.uk or opentable.ca, and numeric restaurant IDs. Search works with the location names OpenTable knows.

Why does Search find fewer restaurants than OpenTable says?

OpenTable's list pages change and overlap while you read them. For big cities the Actor may get fewer unique restaurants than the total OpenTable shows; it says so in a warning. Search a smaller area for a fuller list.

How fresh is the data?

Every row is read live during the run; fetchedAt shows when. Ratings on restaurant rows are OpenTable's averages of recent reviews.

Support

For bugs, questions or feature requests, open an issue on the Issues tab of this Actor. Add the run ID, the input mode and the error text. Never send API tokens, passwords or account data.