Looking for pages
How a website is proofread
List. We read the site's robots.txt file to find the declared sitemap, then try the usual sitemap addresses. Sitemap index files are followed, and you can also enter the address of a sitemap file directly. Without a sitemap, we crawl at most 20 pages by following links, for 18 seconds at most: a deliberately short crawl that will not overload the site.
Select. The filter narrows the list down to addresses containing a word: a section, a year, a product category. “Select all” only selects the visible pages. Beyond the first 500 pages, a button shows the next ones.
Check. Selected pages are read one by one. Their visible text, title and meta description are checked by LanguageTool, in the English variety you chose. Each line then shows its number of mistakes, and the “View” button lists them with their context and the suggested fix.
Why proofread a whole website
Typos on a website cost more than they seem: visitors trust a page less when it is full of mistakes, and a misspelled product name or headline can hurt how people find it. Checking every page at once is much faster than reading them one by one, and it is the best way to catch the mistake that was copied onto twenty pages of a blog or an online shop.
Menu, header and footer
By default, the header, the navigation and the footer are checked along with the rest of the page, since a typo in a menu shows on the whole site. These blocks repeat on every page, though: the same mistake will be counted once for each page you check. Once they are fixed, untick the matching option to see only the mistakes specific to each page.
Choose which mistakes to report
The Spelling, Grammar, Punctuation and Style boxes let you keep only the categories you care about. Untick Style, for example, to see only real mistakes without the writing tips. Your choice is remembered in this browser and also applies to the text checker.
Export the report
The export button creates a CSV file that opens in Excel, LibreOffice or Google Sheets: one row per mistake, with the page address, the category, the excerpt, the explanation and the first suggestion. It is the ideal format to share the corrections among several writers.
Limits
To keep the service free and available to everyone, each IP address can check up to 500 pages and 3,000,000 characters per hour. Pages built entirely with JavaScript, or protected by a password, cannot be read.