Disclosure: I'm Lindenbyte, the small studio that built the Apify Actor used below, and this post was
drafted with AI assistance. The Wayback Machine itself is free and you can do all of this by hand; this is
a bulk shortcut.
An expired or auctioned domain comes with a past. Sometimes that past is a good one (years of a real site,
backlinks, trust). Sometimes the domain dropped, sat dark, got picked up by a casino affiliate or a
replica-watch shop, and dropped again. Registrars and auction lists rarely tell you which. The Internet
Archive usually can.
For one domain, the free route is fine: open https://web.archive.org/web/*/example.com, scroll the
calendar, click a few years. For the 50 names on today's auction list it is an afternoon. This post does the
same check in bulk and gives you one summary row per domain.
1. Put your candidate domains in one run
Open the Wayback Domain History checker and paste
your list. To show what the output looks like I used a deliberately mixed set: a long-lived personal site, a
famous dead service, a sold brand, a social network that went dark, and a name that was never registered.
{ "domains": [ "paulgraham.com", "geocities.com", "mtgox.com", "friendster.com", "pets.com", "lindenbyte-never-registered-48213.com" ], "fetchTitles": true, "maxTitleFetches": 10 } fetchTitles opens up to 10 archived copies of each homepage (spread over the years) and reads the
<title>, which is where most of the useful signal comes from. Leave the archived URL list (includeUrls)
off for due diligence: it is slow on big domains and you don't need it to judge a name.
The run took 4 min 24 s for the six domains. The Actor goes slowly on purpose (the Wayback APIs are free and
rate limited), so for a long list, start it and come back, or schedule it.
2. Read the summary row
Each domain gets one row. The fields that matter most for a buying decision:
| Domain | First seen | Capture days | Longest gap | Flags |
|---|---|---|---|---|
| paulgraham.com | 2000-09-18 | 4,282 | 6.5 months (2001) | none |
| geocities.com | 1996-10-22 | 5,401 | 6.7 months (1997) | redirected-elsewhere |
| mtgox.com | 2007-05-13 | 1,607 | 22.8 months (2007-09 to 2009-08) | none |
| friendster.com | 2002-05-25 | 5,084 | 15.9 months (2020-11 to 2022-03) | redirected-elsewhere |
| pets.com | 1996-12-21 | 1,369 | 19.8 months (1997-98) | redirected-elsewhere |
| lindenbyte-never-registered-48213.com | - | 0 | - | never-archived |
What each part tells you:
-
firstCapture/archiveAgeYears: when the domain first had something on it. paulgraham.com has 26 years of history; that is a different asset from a name first seen last spring. -
capturesPerYearandmissingYears: continuity. "Capture days" means days with at least one homepage capture. Friendster shows 308 capture days in 2020, zero in 2021, 2 in 2022, then 186 in 2023. That hole is exactly the pattern you look for on an expired domain: the old site stopped, something new started later. -
titleSamples: what was actually there. Friendster's titles go from "Friendster - Home" (2006-2009) to "Friendster.com - Living the Game" (2013-2018, the gaming era) to "Friendster" again in 2024 and "Friendster — real friends, real life" in 2026: three different sites on one name. pets.com starts as "Coming Soon!" (1996-97), becomes the famous store, then a community site ("The Online Community for Dog, Cat, Pet Lovers", 2010). -
redirects/externalRedirectDomains: where the homepage pointed. pets.com redirected topetsmart.com(2003, 2021, 2026) andmypetsmart.com(2015); geocities.com to three Yahoo hosts; Friendster in 2020 tomolglobal.mol.com/investor-relations/(its owner at the time). A redirect to another site usually means a sale, a merger, or, on random expired names, a link network. -
statusCodes: paulgraham.com served 200 on 3,947 of 4,282 capture days. geocities.com: 3,317 of its days were 3xx redirects, so the "30-year-old domain" has mostly been a pointer to Yahoo since 1999. - Never archived is an answer, not an error: you get a row with
archived: falseand zero captures.
3. Flag the risky ones automatically
The Actor already sets riskFlags (spam-keywords-in-titles, parked-or-for-sale, language-change,
redirected-elsewhere, long-gap for 24+ months, never-archived, never-served-200). For buying
decisions I prefer stricter rules, so here is a short Python script on top of it (apify-client 3.x):
from apify_client import ApifyClient import os client = ApifyClient(os.environ["APIFY_TOKEN"]) run = client.actor("lindenbyte/wayback-domain-history").call(run_input={ "domains": open("candidates.txt").read().split(), "fetchTitles": True, "maxTitleFetches": 10, }) for d in client.dataset(run.default_dataset_id).iterate_items(): if d.get("error"): print(f"{d['domain']}: lookup failed ({d['error']}), retry later") continue reasons = list(d["riskFlags"]) if d["archived"] and d["longestGapMonths"] >= 12: reasons.append(f"dark {d['longestGapMonths']} months from {d['longestGapFrom'][:7]}") recent_holes = [y for y in d["missingYears"] if y >= 2015] if recent_holes: reasons.append(f"no captures in {recent_holes}") for s in d.get("spamSignals", []): reasons.append(f"'{s['keyword']}' ({s['category']}) in {s['years']}") print(f"{d['domain']}: {'; '.join(reasons) or 'clean'}") On the sample above this prints clean only for paulgraham.com. mtgox.com is flagged for the 22.8-month
dark period (2007-09 to 2009-08, before the exchange launched), friendster.com for the redirect plus the
15.9-month gap and the empty 2021, pets.com for the redirect plus a 19.8-month gap back in 1997-98,
geocities.com for the redirect, and the unregistered name as never-archived.
None of my six domains had spam or parked pages in their titles, which is the point of a known-good test
set. On real drop lists, the spamSignals (casino, pharma, adult, replica, loan keywords) and
parkedOrForSaleYears fields are where most rejections come from. Treat them as signals: open the archived
link of the flagged year and look before deciding.
4. Make it a routine
If you buy from auction lists, export the candidates as text each morning, run the list on a schedule, and
send the dataset to a Google Sheet or webhook. Check the "clean" ones by hand, skip the rest.
Limits (honest ones)
- It reads the archive, not WHOIS. Archive age is when a site existed, not when the name was registered.
- Capture statistics are for the homepage. A domain can have a clean homepage history and spammy inner pages.
- Spam detection is a keyword check on titles and meta descriptions, not a verdict.
- Some archived redirects cannot be resolved (two of Friendster's came back with a fetch error); those years are listed with the error instead of a target.
- Sites excluded from the Internet Archive look the same as never-archived ones.
Cost: $0.004 per domain plus $0.003 per run, so this six-domain check was about $0.03 and a
100-name auction list about $0.40. Lookups that fail are not charged.
Not affiliated with the Internet Archive. If you use the Wayback Machine a lot, consider
donating to them.