How to use the Wayback Machine to recover your content

Updated: 2026-10-27

It’s late, the host’s support chat has closed, and the article you wrote three years ago — the one people still link to — is a 404. You don’t have a copy. You vaguely remember someone saying “it’s probably on the Wayback Machine.”

It probably is. Some of it. Here is how to use the Wayback Machine to find out. It is the Internet Archive’s public archive of web pages, and for a lot of small sites it’s the only copy left. But it isn’t a backup of your site. It’s a library that visited your site now and then, kept what it could carry, and filed it by date. Using it well is mostly about reading that filing system.

How the Wayback Machine works, in one flow

A crawler visits a URL → saves the page as it looked at that moment → stamps it with the date and time → files it under that URL.

Each saved copy is a “capture” (or snapshot). Your site isn’t stored as one thing; it’s stored as many separate URLs, each with its own list of captures from different days. The homepage might have hundreds of captures. A blog post nobody linked to might have none.

That’s the whole mental model. Everything below follows from it.

Step by step: finding your old pages

1. Search by URL, not by words

Go to web.archive.org, type the domain or the exact page URL, and press “Browse History”. The Internet Archive’s help page is clear on what search can and can’t do: Site Search helps you find the homepages of sites by the words people use to describe them, and full-text search across archived pages isn’t available. So if you remember a sentence from your article but not its URL, the Wayback Machine can’t find it from the sentence. Start from the address.

2. Read the calendar

You’ll get a timeline of years and a calendar with dots on the days a capture exists. The colours matter, and the help page spells them out:

Blue — a normal page (HTTP 2xx). Usually what you want.
Green — a redirect (3xx). The page pointed somewhere else that day.
Orange — a client error (4xx), such as a page that was already deleted.
Red — a server error (5xx). The site was broken when the crawler came by.

Pick a blue dot from a date when you know the page was in good shape. If the last few months before your site died are orange and red, go back further. The newest capture is often the worst one.

3. Read the timestamp in the URL

Every archived page has an address like web.archive.org/web/20000229123340/http://example.com/. That long number is the capture time in yyyymmddhhmmss format — this one is 29 February 2000 at 12:33:40. Edit the number and the Wayback Machine jumps to the nearest capture.

To list every capture of a URL, replace the timestamp with an asterisk: web.archive.org/*/www.yoursite.com. To list captured URLs under a whole domain, people commonly add a trailing wildcard to the address, like web.archive.org/*/www.yoursite.com/*, which is the form the Internet Archive’s own Wayback Machine help page gives for seeing the files archived for a site.

4. Save what you find

Once you’ve found a good capture, copy the text out into a document and save the images you need. Do it page by page for a handful of pages. For a whole site, page by page stops being reasonable around the tenth tab — that’s what downloaders are for (step 6).

5. Check the gaps before you plan anything

The help page lists the reasons a page might be missing, and they’re worth knowing before you promise anyone a full restore:

The crawler didn’t know the page existed at the time. Pages with no links pointing to them may never be found.
The site’s robots.txt may have prevented crawling.
Password-protected pages aren’t archived.
Pages that depend on JavaScript or the original server often don’t archive well.
Images show broken or grey when they weren’t archived.
The site owner may have asked to be excluded.

The quick way to see what’s there for a whole domain is to scan a domain in the Wayback Machine with our free scan: it shows the old design, the URL structure and a few sample pages, without downloading the full site.

6. Rebuild a whole site from captures

When you need more than a few pages, there are two honest routes.

Do it yourself. The free open-source Wayback downloader (no AI, runs on your own machine) downloads an archived site as HTML or as a WordPress import file. It works. What it can’t do is judge: raw archives mix captures from different years, so you may get a 2019 header on a 2016 page, missing assets, and the spam pages a hacker left behind. You sort that out by hand.

Have it done. If you’d rather recover deleted website content as a working site, my agent reviews the design and every page: picks the best capture per page, fixes layout breaks, drops junk and thin pages, keeps the old URL structure and flags what’s missing. A human checks each job before delivery. You need to own the rights to the content you’re restoring.

A worked example (hypothetical)

Say you ran a small recipe blog at example.com, and the host deleted the account after a billing failure.

You search example.com and see captures across six years, thick in the middle, thin at the end.
The last two months show red dots — the server was already failing. You go back to a blue capture from the spring before.
The homepage looks right. You click into a recipe: the text is there, two of four photos are grey boxes. Those images weren’t archived.
You list the captured URLs and find your 40 best posts, plus 300 pages in a language you don’t speak, dated the month before the site went down. That’s an old hack, preserved faithfully alongside everything else.
You decide: rebuild the 40 posts, drop the 300 spam pages, and never link to them again.

Step 4 is the one that catches people. The archive keeps what it saw, good and bad. It doesn’t know which is which.

Common mistakes

Restoring the newest capture. It’s often the broken or hacked one. Pick the last good date.
Expecting yesterday’s page. The help page notes a 3–10 hour lag between a crawl and when it shows up — and many small pages are captured rarely, not daily.
Treating Save Page Now as a backup. It saves one specific page, one time. The Internet Archive’s help page on saving pages says it doesn’t save the outlinks, and it doesn’t add the URL to future crawls. Your own backup plugin is still your backup.
Assuming the Archive will package your site for you. Its help page says it no longer packs up lost sites. The data is there to browse; turning it back into a site is on you, or on whoever you hire.
Restoring content you don’t own. Archived doesn’t mean free to reuse. Restore your own work, or work you have the rights to.

For the technical reader: the APIs

The Internet Archive documents two Wayback APIs. The Availability API (archive.org/wayback/available?url=example.com) returns JSON telling you whether a URL is archived and the closest capture, optionally near a given timestamp. The CDX Server API allows complex querying and filtering of capture data, which is what downloaders use to list every capture of every URL on a domain. Be polite with it: the documentation I read gives no request-rate figure, only a cap on results per query, so go slowly and expect throttling on bulk pulls (Internet Archive, CDX server documentation).

If you want your pages removed instead

Site owners can ask the Internet Archive to exclude their site by emailing [email protected] with the URLs, the time period, and when they controlled the site. The Archive says it reviews requests and doesn’t guarantee the outcome in advance.

Takeaway

Search by URL → pick a blue capture from a good date → check what’s missing → copy a few pages by hand, or download the lot and clean it. If the Wayback Machine didn’t capture what you need, there are a few sites like Wayback Machine worth checking, though fewer than most lists claim.

The archive remembers your site better than you do. It just doesn’t remember which version you liked.

Wayback Press is not affiliated with the Internet Archive. The Wayback Machine is the Internet Archive’s service.

Wayback Press (waybackpress.com) is an independent restoration service. Not affiliated with the Internet Archive or the Wayback Machine, nor with the Shift8 open-source “WaybackPress” command-line tool of the same name.