A site is edited or taken offline
A link that used to work now returns a dead page, a "domain for sale" notice, or a redesigned site with the content stripped out. The fix is to stop trying the live URL and go straight to its archived history instead.
- Copy the exact URL of the page that disappeared, including the full path.
- Paste it into the search box at the Wayback Machine.
- Look at the calendar of capture dates that appears above the results.
- Pick a date near when you last knew the page to be working, favoring a bold-circled day, which marks a date with more than one capture.
- Open the snapshot and read the page as it existed on that date, including its images and linked files where those were captured too.
This works because the Wayback Machine indexes captures by exact URL, so a trailing slash, a www. prefix, or an http versus https mismatch can mean the difference between finding a full capture history and finding none at all — worth trying more than one variant of the address before concluding nothing was saved. The Internet Archive's own guidance on using the Wayback Machine walks through the calendar view and how captures are grouped by year, month, and day.
Heritrix and partner crawlers visit URLs on a schedule
Captures exist because a crawler visited that exact URL on that exact day — not because the archive continuously watches every page on the web. The primary tool doing this work is Heritrix, an open-source, archival-quality web crawler built for the Internet Archive and released for other institutions to run as well.
Heritrix and partner crawls follow schedules and priorities that vary by site — some domains get crawled frequently, others rarely, and some pages are never queued at all if nothing ever linked to them. Each visit stores the page's HTML along with its images, stylesheets, and scripts as a single dated capture. That capture is then indexed by URL and by timestamp, which is why the calendar on a Wayback Machine result page shows some days as available and most as blank.
Two consequences follow directly from this:
- A missing date does not mean the page didn't exist then — it means no crawl happened to run that day.
- A capture that looks broken (missing images, unstyled text) usually means the page loaded but some of its supporting files were not captured in the same crawl.
Retrieval works two ways: through the interactive calendar interface, or by writing a direct Wayback URL that embeds the timestamp and target address, useful for linking a specific capture rather than sending someone to search for it themselves.
A page you rely on is still live but may change
If you are citing a page, quoting it, or relying on it as evidence, the safest moment to preserve it is before it changes, not after. Waiting for a source to break before archiving it means depending on whichever crawler happened to visit at whichever moment — an accident of scheduling rather than a plan.
There are three ways to capture a live page on demand instead of waiting for automatic scheduling:
| Method | Where it runs | Best for |
|---|---|---|
| Save Page Now | Wayback Machine website, paste URL and submit | One-off capture of a single page you're viewing right now |
| Browser extension | Chrome Web Store | Saving pages as you browse, without switching tabs |
| Mobile app | iOS and Android app stores | Archiving a page found on a phone, or checking archive status while reading |
Whichever method is used, the result is a permanent, dated URL for that specific capture. Cite that archived address in a bibliography, a legal filing, or a research note instead of the live page's address — the archived link points to the exact content seen on the exact date it was saved, and it survives the source page being edited, paywalled, or deleted afterward. For academic or legal use, the date visible in the Wayback Machine's own timestamp on the capture serves as the citation date, distinct from the date the underlying page was originally published.
A page or date is missing from the archive
An empty calendar for a URL has one of two causes, and they call for different responses. Either no crawler ever visited that exact address, or the page was captured and then excluded from public view after a removal request.
Before concluding the page was never archived, check these possibilities in order:
- Try alternate forms of the same address: with and without
www.,httpversushttps, and with or without a trailing slash — each is indexed separately. - Check whether the page redirected somewhere else at the time; the old and new URLs may have separate, incomplete histories.
- Look at neighboring dates rather than just the one date needed, since a nearby capture may hold the same content.
- If genuinely nothing exists, archive the current live version now using Save Page Now, the extension, or the mobile app, so at least a future record exists.
Site owners and rights holders can request that captured material be excluded from public results, and some material is withheld in response to such requests or to standard web exclusion signals — which is a distinct situation from the service being unavailable generally. Confusing the two is common, since both produce the same symptom: a page that won't load. The Wayback Machine's own usage guidance covers what an excluded or unavailable capture looks like from the viewer's side.
Internet Archive
The Internet Archive is the non-profit organization; the Wayback Machine is one of the services it runs. This distinction matters because the Internet Archive also operates separate collections of digitized books, audio, and software, and searching those calls for different tools than searching page captures. According to its own description, the Internet Archive was founded to build a digital library of internet sites and other cultural artifacts in digital form, and the Wayback Machine is the interface specifically for browsing archived web pages by date.
Keeping the two names separate clarifies why, for instance, a missing book scan and a missing web capture are unrelated problems with unrelated causes: one belongs to the library's book collections, the other to the crawl-based web archive.
Wayback Machine
The Wayback Machine specifically stores and serves dated snapshots of web pages — nothing more. Each snapshot is a capture of one URL at one point in time, retrieved either by browsing the calendar interface at wayback.archive.org or by constructing a direct link that includes the timestamp.
On the question of whether the service is discontinued: outages, temporarily unavailable captures, and gaps for specific sites are often mistaken for the whole service being down or shut down. These are usually the same causes covered above — missing crawls, exclusion requests, or short-term technical outages — rather than any permanent closure. The service continues to operate as the Internet Archive's core tool for viewing archived pages, and its Wikipedia entry documents its ongoing operation and history in detail.
Web crawler (Heritrix)
Heritrix is the crawler that does most of the collecting behind the Wayback Machine's archive. It is open-source software, built specifically for archival-quality, large-scale web crawling rather than for indexing pages for search, and it is maintained and documented publicly by the Internet Archive.
Because Heritrix and its partner crawlers operate on schedules that prioritize some sites over others, coverage across the web is uneven by design rather than by failure — a heavily linked news homepage might be crawled many times a day, while a personal blog might be crawled once every few months or not at all until someone specifically requests it through Save Page Now. Anyone who wants a specific page preserved reliably should not assume Heritrix will get to it in time; using Save Page Now, the browser extension, or the mobile app to trigger an immediate capture is the only way to guarantee a specific date is covered.
Anyone relying on an archived page for a citation, a legal record, or research evidence should now capture the pages currently in use rather than wait for a scheduled crawl to reach them.
