Twenty-three engines here no longer work. Some were shut down on an announced date; some went dark without a word; some still resolve as domains with no search function behind them. Together they show something more useful than a set of obituaries: search engines die in a small number of recognisable ways, and almost none involve losing on the merits.
The underlying economics
An index is not an asset that keeps working. It is a subscription. The crawl must run continuously or the index decays within weeks, and the bill — servers, bandwidth, storage, engineers — arrives every month whether or not the traffic does. So the question that determines an engine's survival is never "are the results good?" It is "who is paying for the crawl, and do they still want to?" Every pattern below is a variation on that question being answered no.
Pattern one: acquired, absorbed, switched off
The commonest ending. A larger company buys the engine, keeps it running for a while as a second brand, and then closes it because two indexes cost more than one and the brands cannibalise each other. AltaVista is the canonical case, and the ownership chain tells the whole story: built at Digital Equipment Corporation and launched in December 1995, passed to Compaq with the DEC acquisition, then to CMGI, then sold to Overture Services in February 2003, then to Yahoo when Yahoo bought Overture that October — and finally shut down by Yahoo on 8 July 2013, seventeen years after being the most technically impressive engine on the web. AllTheWeb, built in Norway out of a doctoral project, followed the identical route through Overture to Yahoo and was closed by Yahoo in April 2011. Teoma, a genuinely independent crawler and ranking system spun out of Rutgers research, was bought by Ask Jeeves in 2001, retired as a brand in 2006, briefly revived in 2010 as a thin front end over Ask's results, and by August 2026 its domain no longer resolves at all. Magellan was bought by Excite in 1996 and shut in 2001. Neeva — which had built its own crawler and index and blended in partner feeds — was acquired by Snowflake in May 2023 and its consumer search engine closed within weeks; the technology went into the acquirer's AI products.
Pattern two: the parent company's decision
Sometimes the engine is fine and the owner has moved on. Disney bought into Infoseek in 1998, took the rest in 1999, folded it into the GO Network, and announced the closure of GO.com and its search engine in January 2001, with roughly four hundred jobs going and a large intangible write-off recorded in its own regulatory filing. DMOZ, the volunteer-edited Open Directory, was never a crawler at all — its data was licensed out to search engines rather than bought in — and it passed by inheritance from Netscape to AOL to Verizon before being closed in March 2017 by an owner with no interest in it. Ask.com is the most recent: an engine that launched in 1997, gave up its own index in 2010, and was finally closed by IAC on 1 May 2026, leaving a single sunset page carrying the line "Every great search must come to an end."
GoTo.com, later Overture, belongs here with a twist. It never built a general web crawler for its consumer product — its own results were the paid listings it auctioned, with unpaid results backfilled from a supplier — and it invented pay-per-click search advertising, demonstrated publicly in February 1998. Yahoo acquired it in 2003 and renamed it Yahoo! Search Marketing in 2005: the brand vanished entirely while the model it invented took over the industry. The goto.com domain now belongs to an unrelated communications company.
Pattern three: the index cost more than the company had
The purest failures, and the rarest, because most engines are sold before they get here. Cuil launched in July 2008 with a large own-built index, an ex-Google founding team and heavy press attention, and went offline in September 2010 with staff reporting they had not been paid that week; an acquisition had reportedly been in progress and fell through. Its patents were later sold to Google. Gigablast ran its own crawler and its own inverted index for roughly twenty-one years, written and maintained largely by one person, never reselling anyone's results — and stopped serving in early April 2023 with no announcement of any kind. A rival independent engine wrote its farewell notice. EuroSeek ran a European-focused crawl in the late 1990s and did not survive its corporate collapse; the domain today serves a submission-based directory with no crawler behind it.
Pattern four: the pivot — the company lives, the search engine does not
Worth separating from a shutdown, because writing "it died" would be wrong in every one of these cases. Northern Light retired its public web search engine in January 2002 and is still trading as an enterprise market-intelligence vendor. You.com discontinued consumer search entirely and now sells web search APIs to companies building AI products — its own crawler and index are still running, just not for the public. LookSmart went through three lives: a human-edited directory from 1996, a genuine own-index engine after acquiring the WiseNut crawler, which it shut down in 2007, and then an advertising business with no consumer search at all.
Pattern five: a whole generation squeezed out at once
The 1990s metasearch engines went as a cohort rather than individually, because the thing that made them useful — many small indexes that barely overlapped — stopped being true. SavvySearch, out of Colorado State University, was absorbed by CNET's Search.com in October 1999. ProFusion, out of the University of Kansas, was acquired in 2000 and folded into other products. Inference Find closed in March 2001. Mamma is now an affiliate-monetised content site with no search function at all. Highway 61 is the instructive one: it demonstrably operated, forwarding queries to five or six named engines, and no founder, company, launch date or shutdown date for it could be established at all. A working service used by real people can vanish so completely that its own existence is only attested by third-party directories.
And the ancestors, which had nowhere to go
Archie catalogued the filenames on public FTP servers from around 1990 — it never indexed web page content, because there was barely any web to index. It was not killed by a competitor; the medium it served was abandoned. Development stopped in the late 1990s, the last legacy server ran at a Polish university until 2023, and a preservation instance launched in 2024 was offline when checked in August 2026. ALIWEB, presented at the first World Wide Web conference in 1994, never crawled anything: site owners wrote their own index files and registered them by hand. That model could not scale past the site owners willing to maintain a file by hand, and full-text crawling superseded it almost at once — though the same author's other contribution, the robots exclusion standard, is still in use across the web.
What the domains do afterwards
One last pattern, and a practical one for anyone typing an old name into an address bar. Almost none of these domains simply go dark. Yahoo still holds the altavista.com and alltheweb.com names. goto.com belongs to an unrelated business. euroseek.com became a hand-submitted directory, mamma.com became affiliate content, and infoseek.com redirects to a corporate landing page with no search box. teoma.com is the outlier that does not resolve at all. A domain that still loads is not evidence that a search engine still exists, and that confusion is why the engines on this page keep appearing in lists of engines you can supposedly still use.