Paste a Deliveroo store URL - or a bare restaurant id - and get that restaurant's reviews back as rows: who wrote it, the rating and the scale it sits on, the review text, the date, and whether they would order again. One row per review, in the sort order you choose.
Find the restaurant on Deliveroo, copy the address from the bar - or use the id if you already keep one - and choose how you want the reviews ordered.
One row per review, tagged with the query it came from. Deliveroo sits behind Cloudflare, so live runs need a residential proxy.
rating arrives next to max_rating, so the number never travels without the thing that gives it meaning. That matters the moment these rows sit in a table beside another platform's scores.
Choose highest rated, lowest rated, newest or oldest before the run starts. If you only want the complaints, take the lowest-rated slice and stop - you pay for the rows you pull, not for the whole history.
Deliveroo asks reviewers whether they would order again, and that answer arrives as order_again rather than buried in prose. It is the closest thing to an intent column this dataset has.
What a Deliveroo review carries: who left it, how they scored it, what they wrote, when, and whether they would come back.
Two things we would rather state than let you assume. The first: the table below names the eight columns and says what each field is for - and it stops there. It does not tell you what scale max_rating reports, what vocabulary order_again uses, or what format date arrives in. Those are answers a populated export gives, and every run export we hold came back with a header row and no data rows. Publishing a confident guess in their place would be worse than publishing nothing - and in the case of max_rating it would be guessing at the exact thing the column exists to tell you.
The second: Deliveroo sits behind Cloudflare. Live results need a residential proxy, and that is why the exports we hold are empty rather than because the column set is uncertain - the column set is corroborated twice over. If you do not have a residential proxy, this is the page to read carefully before you sign up rather than after.
Run a small free-tier job against a restaurant you know. The header row and the first few rows of that file are the definitive answer for the stores you actually care about.
The export's header row, in order. Descriptions say what each field is for; they deliberately claim nothing about its format, scale or vocabulary - see the note above.
max_rating rather than against an assumed scale.rating is expressed on. The export carries it so you do not have to assume one - which is also why this page does not print a number here.The names are solid; the contents are not yet. The eight columns above and their order are corroborated by two independent sources - the XLSX header row of three separate run exports and the published column array - so build your schema around those names with confidence. What this page will not tell you is what the cells hold: every export we hold is a header with no rows beneath it, so there is no honest statement to make about the scale in max_rating, the vocabulary in order_again or the format of date. And Deliveroo sits behind Cloudflare, so a live run needs a residential proxy - that is the reason those exports are empty, not any doubt about the column set.
Point it at the menu page you would otherwise scroll, or hand over the restaurant id if you already keep one. Sort order and a per-query limit are chosen before the run starts.
A few examples of what delivery reviews are worth once they are rows rather than an endless scroll.
Take the lowest-rated slice for your own stores and read the text column as a list rather than a feed. Late, cold, missing items and wrong order are different problems with different fixes, and you cannot tell which one you have from an average.
Pull the same review set for every competitor in your delivery radius and put the ratings side by side with their scale. The gap between your score and theirs is a number you can act on; a screenshot of their menu page is not.
Sort by newest, re-run on a schedule and keep the rows. A new packaging supplier or a kitchen change shows up in the review text long before it shows up in the average, which moves slowly by design.
No subscription, no minimum, no recurring bill. Your first 500 rows are on us - after that, pay-as-you-go at the same flat rate as every other scraper here.
Every new account, one-time. No credit card required. Per-query limits, sort order, file upload and every export format included.
Roughly $2 per 1,000 reviews. The pre-flight estimator shows the row count and credit cost before a run starts - no surprise bills, no compute units to translate.
Volume pricing, dedicated workers and an SLA for tracking many restaurants on a schedule. Tell us your numbers and we will quote.
LIVESCRAPER10 at checkout.One platform's reviews are one platform's customers. These two widen the picture.
The same job on the Dutch delivery platform. If you operate across both markets, this is the other half of the same question - what the delivery customer said.
The reviews from people who walked in rather than ordered delivery. A restaurant often scores very differently across the two, and the gap is usually the interesting part.
Where people complain about the delivery brand rather than the restaurant. Useful when you are trying to separate a kitchen problem from a platform problem.
Short answer: these reviews are published on a public restaurant page - but they are written by identifiable people, and that is the part worth thinking about.
Everything collected here is shown on a public Deliveroo store page to any visitor, signed in or not: the reviewer as the site displays them, the rating and its scale, the review text, the date, and the would-order-again answer. No login, no paywall, nothing behind an account. There are no email addresses, phone numbers or delivery addresses anywhere in the export, and no column for one.
The author column is personal data. A reviewer's display name plus what they wrote and when identifies a person, and in the EU and UK that brings the export under the GDPR with you as the controller. Reading reviews in aggregate to fix your kitchen is a very different act from building a profile of a named individual, and only the first is what this tool is for. Keep it to the aggregate, and if you do store the rows, have a lawful basis and honour erasure requests.
One practical note that is technical rather than legal: Deliveroo sits behind Cloudflare, so live results need a residential proxy. And as always, Deliveroo's own terms govern automated collection from their site - read them before you scale, rather than relying on the fact that the pages are public. We run no third-party trackers on the data layer, and your exports auto-delete after 30 days.
The questions we hear most. Anything else? Talk to us - humans, not bots, write the answers.
Your first 500 rows are free - no card, no subscription. After that it is $0.002 per row, flat.
Every restaurant on Deliveroo carries reviews from the people who ordered from it, and they sit on a public store page where anyone can read them one at a time. The Deliveroo Reviews Scraper turns that page into rows. You submit store or menu URLs - or bare restaurant ids, one per line, or as a CSV, XLSX, TXT or Parquet upload - pick a sort order and a per-query limit, and each review comes back as a record carrying the reviewer, the rating and the scale it is expressed on, the review text, the date and the would-order-again answer.
The sort order is the part that decides what a run costs. You can take the highest rated, the lowest rated, the newest or the oldest, and because billing is per row, a narrow slice of the complaints costs a fraction of the whole history. For an operations question that is usually the right shape: the lowest-rated reviews for your own stores, read as a list, tell you whether you have a late problem, a cold-food problem or a missing-items problem - which an average never will.
Two honest notes before you start. Deliveroo sits behind Cloudflare, so live results need a residential proxy; that is stated up front rather than left to be discovered. And the data dictionary on this page names the eight columns and says what each is for, but claims nothing about their formats, scales or vocabularies. Every run export we hold came back as a header row with no data rows, for the proxy reason above, so there is no honest thing to say about what max_rating reports or what values order_again takes. The column names themselves are corroborated twice over and can be relied on.
Take a small free-tier run against a restaurant you know and the header row and first few rows are the definitive answer for the stores you care about. Your first 500 rows cost nothing and need no credit card, and after that it is $0.002 per row, flat.