How to download a website for offline use

Press Ctrl+S (Cmd+S on a Mac), choose "Webpage, Complete", and Chrome writes an HTML file plus a folder of assets. On a static page from a decade ago this is all you need.

On a current page it tends to lose CSS background images, lazy-loaded pictures below the fold, and webfonts. You will not find out which until you open the copy offline. Worth trying first precisely because it costs ten seconds; do not build an archive on it.

An extension has an advantage no external tool can match: it sees the page after JavaScript has run, images have resolved and the stylesheet has been applied. That is why saving a web page with one usually produces a copy that looks like the original rather than an approximation of it.

With Site Downloader, one click saves the current tab. Switching to Multi-page turns it into a full site downloader: pick how many internal-link levels to follow and everything lands in a single .zip.

The limits are worth stating plainly. It runs only where your browser runs, it cannot be scripted, and it fetches without your cookies, so nothing behind a login comes down.

HTTrack has mirrored websites since the late nineties and remains free software under the GPL. You point it at a URL, walk a wizard through scan rules and filters, and it recursively pulls the site to a local directory with links rewritten.

What you get for the dated interface is control: include and exclude patterns, bandwidth limits, resumable jobs, mirror updates. What you do not get is a browser. HTTrack fetches raw HTTP, so any page that assembles itself with JavaScript arrives as an empty shell. People searching for an HTTrack alternative are almost always running into exactly that.

On macOS or Linux, the canonical mirror command is:

wget --mirror --convert-links --adjust-extension --page-requisites --no-parent https://example.com/docs/

--convert-links rewrites references for local browsing, --page-requisites pulls the CSS and images a page needs, and --no-parent stops it climbing above the directory you named. Add --wait=1 to be polite to the server.

This is the right tool when the job needs to run unattended, on a schedule, on a machine with no screen. It is the wrong tool for a modern single-page app, and it will not respect a session you are logged into unless you go to some length with cookie files.

Services that download a site for you need nothing installed, which makes them genuinely useful on a locked-down work machine. The price is that a company you have never heard of now has a log entry for every URL you archived, and free tiers ration pages, speed, or both.

They also fetch from their own servers, so anything geo-restricted, rate-limited or browser-gated on the origin site tends to come back wrong.

Unstyled text on a white background

Either the stylesheet never came down, or nobody rewrote its path. Open the HTML in a text editor and look at the <link rel="stylesheet"> tag. If it still points at https://example.com, the tool did not convert it. wget needs --convert-links; a browser extension should have done it for you.

Images missing below the fold

Lazy loading. They were never requested, so they were never saved. Scroll the whole page to the bottom before you download. That forces the browser to load them, and an extension will then pick them up.

The crawl grabbed thousands of pages

Depth was too high, or the site has a calendar, a tag cloud or paginated filters that generate near-infinite URLs. Drop a level and start from a deeper page.

Everything is a login screen

The content requires a session. Tools that fetch without your cookies (most of them, for good privacy reasons) will only ever see the logged-out version.

Saving pages for personal reading, research or a backup is ordinary use of the web; your browser already caches every page you visit. Republishing what you copied, or passing someone else's site off as your own, is a different matter. Copyright governs that, not the tool you reached for.

Two practical courtesies: do not hammer a small server with a deep crawl at full speed, and read the site's terms before archiving at scale or for anything commercial. Treat this as general information rather than legal advice.