How to Scrape Amazon Search Results to CSV: Prices, Ratings, Reviews, and ASINs

With the built-in Shop::Amazon scraper, you can scrape Amazon search results and export product titles, ASINs, URLs, current and strikethrough prices, currencies, ratings, and review counts to CSV.

From an Amazon Search Query to a Ready-to-Use CSV

This guide shows you how to configure a task for one search query, validate the CSV, and save the configuration for future runs. It also breaks down a real test run with 275 Amazon listings.

What Amazon Data You Can Save to CSV

The built-in Shop::Amazon scraper takes a standard Amazon search query and extracts data from the product listings in the results.

For your first export, you only need eight product fields:

The scraper also returns optional data: item.boughtCount contains the displayed number of purchases in the past month, item.additionalinfo contains badges such as Best Seller and Amazon’s Choice, item.sellerscount contains the number of sellers, and item.imagelink contains the product image URL.

How to Set Up Amazon Scraping in A-Parser

What You Need

• A-Parser Pro or Enterprise;
• one or more search queries;
• proxies with IP addresses in the target country are recommended for recurring runs.

Step 1. Select Shop::Amazon

Open the Task Editor and select Shop::Amazon from the scraper list.

Step 2. Add Search Queries

Enter queries in the Queries field just as you would enter them in the Amazon search bar:

wireless headphones
gaming mouse
mechanical keyboard

One to five queries are enough for the first test. You can also expand queries automatically and use letter ranges such as {az:a:z}, but verify the basic task before scaling up.

Step 3. Choose the Amazon Domain and Number of Pages

In NextGen, choose a country such as United States from the Amazon domain dropdown. For an API request, specify the corresponding Amazon domain.

• domain — the target site domain for an API request, such as www.amazon.com, www.amazon.de, or www.amazon.co.uk;
• pagecount — the number of pages to scrape. Set it to 1 for the first test, then increase it to 3–5.

Choose the country from the list in the UI. Through the API, pass the full domain, such as www.amazon.com. A raw value such as germany is not converted to a domain automatically.

Step 4. Configure Proxies and Geolocation

You can run a few test queries from your current IP address. For consistent USD pricing on amazon.com, use a US exit IP address.

If a proxy uses the user:pass@host:port format, enable Use proxy authorization in Proxy Checker. Otherwise, the proxy server may return 407 Proxy Authentication Required.

Step 5. Save the Results to CSV

You do not need to write the template yourself. In NextGen, the CSV template belongs in the scraper’s Result format field, while the ready-made preset at the end of this guide already includes the table header. In the classic UI, use Initial text for the header.

CSV header

In the ready-made preset, the header is stored in resultsPrepend and already contains this line:

Keyword,ASIN,Product Title,Current Price,Old Price,Currency,Rating,Reviews,Product URL

A newline is required after Product URL; otherwise, the first product row will merge with the header. It is already included in the preset. When configuring the classic UI manually, press Enter after the header.

Product rows

Paste this into the Result format field:

[% FOREACH item IN p1.products;
    # If ASIN is absent from the dedicated field, extract it from the URL
    asin = item.asin ? item.asin : item.link.match('/(?:dp|gp/product)/([A-Z0-9]{10})').0;

    tools.CSVline(
        query,
        asin,
        item.title,
        item.price,
        item.oldprice,
        item.currency,
        item.rating,
        item.commentscount,
        item.link
    );
END %]

Here, p1 is the short name assigned to Shop::Amazon, while p1.products is the list of products found. If this is the only scraper in the task, you do not need to change p1.

Why use tools.CSVline? Amazon titles may contain commas, quotation marks, and line breaks. In the test dataset, 199 of 275 titles contained a comma. The function escapes these values automatically so that a title does not split across multiple columns.

File name and task launch

Choose to save the results to a file and enter:

results/amazon-$datefile.format().csv

A timestamp is added to the file name, so a new export does not overwrite the previous one.

If you plan to open the CSV in Excel, enable Write BOM in the output settings. This option is already enabled in the ready-made preset.

Click Add Task. When the status changes to Completed, download the CSV from the task queue.

How to Validate the CSV Before Scaling Up

Check the first 20 rows before scaling up the number of queries and pages:

• prices are stored as decimal numbers, such as 29.99;
• empty prices correspond to listings with See options;
• ratings are stored as numbers, such as 4.5;
• review counts are preserved in Amazon’s abbreviated format: 535, 2.8K, 225.6K;
• each ASIN contains 10 alphanumeric characters;
• every row contains the same number of columns;
• decode HTML entities such as & before analysis if necessary.

Important: an empty price does not always indicate an error. A listing may offer several colors, sizes, or configurations, and Amazon may display the price only after a variant is selected.

Real-World Test: 275 Amazon Listings with a Limit of 30 Pages

One page is enough to validate the CSV format, but not to show whether the task remains reliable over a longer run. We therefore set the same configuration to a maximum of 30 pages and checked prices, ratings, reviews, duplicate ASINs, and missing values. Amazon returned No more pages after page 20, so the run ended before reaching the configured limit.

How the Data Was Collected

We normalized the values before calculating the metrics. Abbreviated review counts were converted to numbers: 2.8K → 2,800 and 225.6K → 225,600. Empty values were excluded from the relevant price, rating, and review-count calculations. Duplicate rows were identified by ASIN.

Running the same task again may return a different set of products because this dataset is a snapshot of Amazon search results from that run.

Results

Amazon search results dataset, collected on September 25, 2026.

What the Results Show

1. Products below $50 dominate the search results. This range contained 203 of the 269 listings with a price, or 75.5%. The average price ($50.67) is higher than the median ($26.99) because it is pulled upward by 28 listings priced at $100 or more.
2. oldprice does not show price history. One product had a current price of $23.98 and a strikethrough price of $299.99, a difference of 92%. This is a reference price for comparison, not proof that the product previously sold at that price.
3. Review counts must be converted to numbers. The K suffix appeared in 181 of 274 non-empty values, or 66%. These strings cannot be compared directly as numbers. After converting 2.8K to 2,800 and normalizing the other abbreviated values, the median was 2,800 reviews.
4. Duplicates should be removed by ASIN. The 275 rows contained 271 unique ASINs: four products appeared twice.

Download the CSV with 275 Amazon listings used for these calculations.

What to Do If the Output Looks Wrong

When A-Parser Makes More Sense Than a Custom Script

A Python script makes sense when you need custom logic and complete control over every request. For recurring scrapes, however, you must separately maintain selectors, proxies, retry logic for network errors, output formatting, and scheduling.

A-Parser combines these components in one task. You can save the configuration as a preset, run it with other queries, and schedule it through Task Scheduler. In Enterprise, you can also integrate the task with your own system through the HTTP/JSON API.

Frequently Asked Questions About Scraping Amazon to CSV

Can I Export Amazon Products Without Writing Code?

Yes. Shop::Amazon is a built-in module, so you do not need to build a custom scraper. Select the module, enter your queries, and copy the ready-made CSV template. You only need to edit the template if you want a different set of columns.

Do I Need Proxies to Scrape Amazon?

You can use your current IP address for a small test. For recurring runs and consistent regional results, rotating proxies in the target country are preferable. The exit IP location affects the currency and localization of Amazon results.

How Do I Avoid Duplicate Rows in the CSV?

Use ASIN as the unique key. The same product may appear on several pages or for different queries. For a combined catalog, remove duplicate ASINs; for rank tracking, keep each query + ASIN pair unique.

Resources and Next Steps

• Official Shop::Amazon documentation
• Download the dataset of 275 Amazon listings
• Choose A-Parser Pro or Enterprise
• Get A-Parser Premium proxies

Ready-to-Import Preset

In the Task Editor, click the import button to the right of the Task preset field and select Import preset. Expand the spoiler and paste the Base64 code into the import field.
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For the first run, keep pagecount = 1, replace the test query with your own, and inspect the first 20 CSV rows. If the prices, currency, and columns look correct, increase the number of pages.

← View all published scraping guides