Downloading results

Every results page has buttons that download what the search found: URLs, CSV and CSV+snippets. A signed-in account also has Cluster, which saves the results instead of downloading them. Rows come in the same order as on the results page.

What each button gives

The file is named after the query with everything but Latin letters, digits, - and _ left out: "static.hotjar.com" gives statichotjarcom.csv.

The format

• UTF-8. The first line of CSV and CSV+snippets names the columns: url;rank or url;rank;snippets. The URLs file has no such line.
• Columns are separated by a semicolon ;. A cell that contains a semicolon, a comma, a tab, a double quote or a line break is put in double quotes, and a quote inside it is doubled. This is standard CSV: a spreadsheet or a CSV library reads it as is, with the semicolon chosen as the separator, and a cell stays whole even if the comma or the tab is chosen too.
• Rank is the site's place in the popularity ranking. It is empty for a site that has no rank.
• All snippets of one page are in one cell, joined with ....
• A snippet that starts with =, +, - or @ gets an apostrophe in front, so a spreadsheet shows it instead of running it as a formula. Numbers such as phone numbers are left as they are.

How many rows

Rows are taken from the top of the results, up to the plan's limit: Results / each search or download for URLs and CSV, Download with snippets (rows) for CSV+snippets. The numbers are on the Pricing page. Without signing in, a file holds the first page of results: up to 20 rows.

Websites hidden on your plan - the rows shown as Sign up to view or Upgrade to view - are not in the file. Results go by popularity and the hidden websites come after the open ones, so a file can be shorter than the limit, or have no rows at all; then the download buttons are inactive.

The hint on each button says how many rows the file will have, for example 7 of 1,059 rows: websites hidden on your plan are not downloaded. When the page does not show where the open rows end, it says At most: the file stops at the first website hidden on your plan, which may come sooner.

When the file is shorter than the results, its last line says why:

# Truncated at the plan limit: 1000 of 22694 results.
# 7 of 1059 results: websites hidden on your plan are left out.

The first form means the plan's row limit was reached, the second that the websites hidden on your plan were left out. The line starts with # and is written in the language of the site you downloaded from; its numbers have no thousands separator. It is not a result: skip it when reading the file into a program. When it is not there, the file holds all the results.

A download is charged like the search itself: one search from the daily allowance, and CSV+snippets also one download with snippets - see quotas and rate limits. When the day's searches are used up, a download carries what the free plan allows, and the last line says the file was cut.

From a script

The buttons are links on the results page; the address in them works for ten minutes from the network address that opened the page. To download from your own code, use the API: /v1/search with format=csv or format=txt. The old export urls with ?key= keep their old format exactly: cells are not quoted, there is no last line, and a site without a rank shows >30M.

Other downloads

• A cluster - the download icon in the list of clusters gives a text file of its domains, one per line.
• Extraction - a CSV separated by commas, not semicolons, with a header row: the domain, then one column per expression. Several values found on one site share a cell, one per line. The last line, starting with #, says how many extraction points the run used; when the day's points run out, the line before it says where to continue. See Extraction.