Wayback Machine Downloader

Example

What a restore actually returns

Two real runs against dead dot-com era sites, with the numbers exactly as they came out and screenshots of the recovered pages.

  1. Read the indexEvery URL the Archive holds
  2. Fetch each captureAs recorded, no toolbar
  3. Rewrite the linksPoint at the local copy
  4. Write to diskOriginal folder structure
The whole method. Everything it reads is public.

kozmo.com, restored

Kozmo promised free delivery in under an hour and folded in 2001. The Archive kept its homepage, its stylesheets and most of its images, so the recovered copy opens looking very close to the original.

The recovered kozmo.com homepage rendering in a browser: an orange page offering free delivery in under an hour, panels for first-time users and members, and a list of categories including 25,000 video rental titles and Dreamcast game rentals.
The recovered homepage, opened from local files with no internet connection to kozmo.com. Two button images were never captured and show as placeholders; everything else is what the site served.
Pages written
895
Failed
5
Images
192
Stylesheets
383
Time
6m 35s
On disk
38 MB

theglobe.com, and what thin coverage looks like

theglobe.com was one of the first social networks and the biggest IPO pop of the dot-com boom. Its pages survive but most of its images do not, which is the more common outcome and worth seeing before you decide anything.

The recovered theglobe.com homepage from January 1999. The layout, navigation and headlines about Michael Jordan retiring and the Clinton impeachment case all render, but most images show as broken placeholders.
January 1999. Michael Jordan retiring, the impeachment case opening, a chat with Richard Grieco promoted for the following Monday. The structure, text and links come back; most images were never captured and no tool can return them.

This is the honest range. Some sites come back nearly whole and some come back as structure and text. The Archive decided which, years ago, and the free coverage check tells you which one you are looking at before you pay.

About those failures

Five URLs failed on the Kozmo run and one on theglobe. In both cases they were malformed URLs that spammers had injected into the Archive’s index for the domain, which the Archive itself rejects. That happens on most older domains. It is data quality at the source, not a fault in the restore.

Try it against your own domain

The coverage check is free, takes a few seconds and needs no email. It reads the same index these runs used and tells you which years hold captures for your domain and where the gaps are.