AltaVista

AltaVista is gone, and has been since 2013

AltaVista is not a search engine you can use. Yahoo announced its closure on 28 June 2013 and shut it down on 8 July 2013, pointing the domain at Yahoo Search. There is no AltaVista index, no AltaVista crawler, and no AltaVista result set anywhere on the web today. That is what the Discontinued verdict at the top of this page means, and it is not a soft judgment about a service in decline — the product does not exist.

The closure was less dramatic than it sounds, because by 2013 AltaVista had already stopped being an independent engine. After Overture bought it in February 2003 and Yahoo bought Overture that October, AltaVista's own index was retired and the site displayed Yahoo's results behind AltaVista branding. The 2013 shutdown closed a front-end. The engine underneath it had been switched off years earlier — exactly when is not documented in any primary source we could locate, so this page does not put a date on it.

This matters for anyone compiling a list of Google alternatives. AltaVista appears on a surprising number of such lists, usually because the writer typed the address, watched the browser load a working search box, and assumed the service was alive. It is not. What loads is Yahoo.

Why altavista.com still seems to work

Checked on 19 August 2026, altavista.com returns an HTTP 302 redirect to search.yahoo.com/?fr=altavista. The redirect Yahoo installed in 2013 is still running thirteen years later, and the fr=altavista referral parameter is still attached, so Yahoo is still counting the traffic that arrives from the old brand.

Three things are worth separating here, because conflating them is the single most common error about AltaVista:

• The domain resolves. Someone still renews it and a web server still answers on it.
• The redirect works. You land on a functioning search page and get results for your query.
• The engine does not exist. Those results are Yahoo's — and Yahoo has no crawler of its own either. Yahoo's web results come from Microsoft Bing under a long-running syndication arrangement.

So typing altavista.com in 2026 puts three brand names between you and one index, and none of the three is AltaVista's. A domain that resolves is not a search engine that works. If a page tells you AltaVista is still operating, that page is describing a redirect.

How it worked: Scooter, and an index that held the whole page

AltaVista's technical achievement was not its ranking. It was its crawler and the size of the index (the engine's own stored copy of the web) that the crawler could fill.

The crawler was called Scooter and was written by Louis Monier at Digital Equipment Corporation's Western Research Laboratory in Palo Alto. Its distinguishing property was concurrency: Scooter was multi-threaded, keeping large numbers of HTTP fetches in flight at once instead of requesting one page, waiting, and requesting the next. That is a mundane-sounding design decision with an enormous consequence — it let one team build an index an order of magnitude larger than its contemporaries could. This characterisation is well established in the secondary literature rather than drawn from a surviving DEC technical report, and the specific concurrent-connection figures that circulate online are unverified, so no number appears here.

The indexer, built substantially by Michael Burrows, stored the full text of every page rather than titles, keywords and metadata alone. Predecessors indexed descriptions of documents; AltaVista indexed the documents. Query handling was weighted Boolean, with basic and advanced modes, and later natural-language queries. From December 1997 it also hosted Babel Fish, a free machine-translation service built on Systran and one of the first of its kind on the web.

Independent snapshots of how big that index was are rare, and the ones that survive are engine-reported rather than audited. A research paper by Gerald Vorndran of Kent State University, dated August 1997 and archived by ERIC at the US Department of Education as document ED 413 900, compiled the figures the engines themselves were claiming at that moment: more than 30 million URLs for AltaVista, more than 50 million each for Excite and HotBot, and 30 to 50 million for Infoseek. Those are the engines' own numbers as the author collected them, not measurements anyone checked. The paper's verdict on the field is the more striking part. Writing for an audience of professional searchers, Vorndran concluded that "The Web search interfaces, as a whole, still trail the DIALOG search interface in terms of the quality, quantity, depth (robustness), and usability of the search system" — in 1997, the people who searched for a living did not yet regard any web engine, AltaVista included, as a serious retrieval tool.

The scale Scooter eventually reached is documented in one of the very few published studies built on AltaVista's own production crawl data. In Graph Structure in the Web, presented at the Ninth International World Wide Web Conference in 2000, Broder, Kumar, Maghoul, Raghavan, Rajagopalan, Stata, Tomkins and Wiener — a team drawn from AltaVista, IBM Almaden and Compaq — worked from an AltaVista crawl of May 1999 covering 203 million URLs and 1,466 million links, and a second crawl from October 1999 covering 271 million URLs and 2,130 million links. Those are the raw dimensions of what one company's crawler was holding at the end of the 1990s, recorded by the people who ran it.

The paper is better remembered for what it found inside that data. The 1999 web had a strongly connected core — a region in which every page could reach every other by following links — of 56,463,993 pages, wrapped in much larger regions that fed into the core or led away from it, the shape the authors called a bow tie. Between two randomly chosen pages a directed path existed only about 24 percent of the time, and where one existed it ran to roughly 16 hops. This is a 1999 snapshot and nothing more. The web's link structure has changed substantially since, and none of these figures describes the web as it is now.

What AltaVista never had was link-graph ranking. It scored pages on the words in them. Google's PageRank scored pages on which other pages pointed at them, and on a web that was rapidly filling with keyword-stuffed junk, that difference decided the market. Anyone who says AltaVista's innovation was its algorithm has the story backwards: the algorithm was its weakness.

Size was never what AltaVista lost on, and a contemporaneous table makes the point. A survey presentation preserved in the CS748T course archive at the University of Waterloo's David R. Cheriton School of Computer Science records late-1990s index sizes side by side: AltaVista at 140 million pages, HotBot at 110 million, Northern Light at 67 million, and Google — then still running at google.stanford.edu — at 25 million, with a citation graph of 518 million links behind it. The deck also notes that the fastest crawlers of the period could traverse up to 10 million pages a day. These are secondary compilations inside an academic deck rather than audited measurements, so they are worth citing as what the deck recorded rather than as counts anyone verified. The ordering is the part that matters: at that moment Google held the smallest index of the four.

There is a neat irony in its origins. AltaVista was conceived by Paul Flaherty at DEC as a demonstration of the company's 64-bit AlphaServer hardware, which had enough memory to hold a web-scale index in RAM. The most important search engine of the mid-1990s began life as a hardware marketing exercise.

Who owned it, in order

The ownership chain explains most of what happened to the product:

1. Digital Equipment Corporation (1995–1998) — built it, launched it on 15 December 1995 at altavista.digital.com, treated it as a showcase rather than a business.
2. Compaq (from June 1998) — inherited AltaVista by acquiring DEC, and bought the altavista.com domain from an unrelated software firm, AltaVista Technology Incorporated. The purchase price is widely reported as US$3.3 million; that figure is repeated everywhere and primary-sourced nowhere, so treat it as reported rather than confirmed.
3. CMGI (majority stake, June 1999) — filed for an IPO that was abandoned when the dot-com market collapsed, and rebuilt the site as a portal.
4. Overture Services (February 2003) — acquired it for approximately US$140 million in cash and stock.
5. Yahoo (October 2003) — acquired Overture and with it AltaVista, in the same year it acquired Inktomi. Yahoo has held the domain since; Yahoo itself has been owned by Apollo Global Management since September 2021.

Note who is not on that list: Google. Google has never owned AltaVista, never bought its index, and never acquired any part of it.

How it made money, and how big it got

AltaVista was free to use and funded by advertising: banner display and sponsorship at first, later supplemented by paid-placement and paid-inclusion listings once Overture, a company built on paid search, owned it. There was also a licensing business, AltaVista Search Intranet, which sold the engine to companies for searching their internal networks.

The strategic error is easy to identify with hindsight and was visible at the time. Under CMGI between 1999 and 2002 AltaVista was reworked into a general-purpose portal — email, shopping, a directory, news — chasing Yahoo's advertising model, because portals were what the market rewarded. It did this in precisely the years when relevance ranking became the competitive battleground. A company with the best crawler on the web spent the PageRank era redesigning its home page. Search quality and focus declined; Google's did not.

The audience numbers, with the caveat they need

No modern-methodology market-share series covers AltaVista's peak. StatCounter did not exist in the relevant period and no comparable audited panel data survives, so any figure claiming AltaVista's "market share" in the 1990s should be treated with suspicion. The numbers that do exist are reported claims, mostly compiled from contemporaneous press:

• 15 December 1995: around 300,000 hits on the first day.
• Late 1997: more than 80 million hits per day.
• 1997: approximately US$50 million in sponsorship revenue.
• 1998 and 2000: ranked the eleventh most-visited website.
• 2000: used by 17.7% of internet users.
• February 1998: 45% of professional researchers named it their preferred engine.

The last of those is the most telling. Among people whose job was finding things, AltaVista was the default choice barely five years before it was sold for scrap value relative to its peak. Search markets move faster than brand loyalty does.

One day of real queries, 8 September 2002

One AltaVista usage dataset is not a reported claim. Jansen, Spink and Pedersen, writing in the Journal of the American Society for Information Science and Technology in 2005, analyzed a 24-hour AltaVista transaction log — a record of the queries the engine actually received — supplied by the company and covering 8 September 2002. It held 1,073,388 queries from 369,350 sessions. The average query ran to 2.92 terms, 20.4 percent of queries were a single term, and 72.8 percent of users viewed only the first page of results.

That last number is the dated primary source behind a claim usually repeated without one: that almost nobody looks past page one. It measures AltaVista on a single day in September 2002 and nothing else. It is not a statement about how people use any engine now, and the log it came from was never published and cannot be regenerated.

Timeline

• 15 December 1995 — public launch as altavista.digital.com from DEC's Palo Alto labs.
• 1996 — Yahoo contracts AltaVista to supply the web results behind its human-edited directory.
• December 1997 — Babel Fish launches, offering free machine translation on the web.
• June 1998 — Compaq completes its acquisition of DEC and buys the altavista.com domain.
• June 1999 — CMGI takes a majority stake; the planned IPO is later abandoned.
• 1999–2001 — repositioned as a portal as Google's PageRank ranking takes the technical lead.
• February 2003 — Overture Services acquires AltaVista for roughly US$140 million.
• October 2003 — Yahoo completes its acquisition of Overture.
• 2011 — Yahoo lists AltaVista among properties to be sunset, then defers.
• 28 June 2013 — Yahoo announces the closure, as reported by The Register.
• 8 July 2013 — the site goes dark and redirects to Yahoo Search.

Common misconceptions

"AltaVista was the first search engine." It was not. WebCrawler in 1994, Lycos in 1994 and Infoseek in 1994–95 all preceded it. Its genuine claim is narrower and more interesting: the first fast, full-text, web-scale index behind a simple public search box.

"AltaVista was a Silicon Valley startup." No. It was an internal research project at Digital Equipment Corporation, a Massachusetts minicomputer maker, developed at DEC's Palo Alto laboratories to sell AlphaServer hardware. It had no founding-startup structure, no venture round, and no independent existence until it was already an established product.

"Google bought AltaVista." Google never owned it. DEC, then Compaq, then CMGI, then Overture, then Yahoo.

"AltaVista still works." The domain answers; the engine does not. See the 302 redirect described above, verified on 19 August 2026.

"Yahoo killed a working engine in 2013." Yahoo closed a front-end that had been showing Yahoo's own results for years. The independent engine died quietly, well before the announcement, and the obituaries in July 2013 were for the brand.

What AltaVista is good for now

As a live tool: nothing. There is no version of AltaVista that returns AltaVista results, no archived index that can be queried, and no successor product carrying the technology forward. A reader looking for an engine that is not Google gains nothing here — the redirect delivers Yahoo, and Yahoo's web results come from Microsoft Bing, so the destination is one of the two large Western indexes rather than an alternative to them.

As history, it is genuinely important, and the useful lesson is structural rather than nostalgic. AltaVista proved that indexing the entire text of the entire web was tractable, two and a half years before Google launched, and it lost anyway — because indexing everything and ranking it well are separate problems, and the second one turned out to be the business. Every engine covered on this site is still working on that distinction.

Frequently asked questions

Sources

• Broder, Kumar, Maghoul, Raghavan, Rajagopalan, Stata, Tomkins and Wiener; AltaVista, IBM Almaden and Compaq — Graph Structure in the Web, Proceeding… (2000, PDF)
• Bernard J. Jansen, Amanda Spink and Jan Pedersen — A Temporal Comparison of AltaVista Web Searching, Journal of the American Society for Information… (2005, PDF)
• Gerald Vorndran, Kent State University, archived by ERIC, US Department of Education — A Search Engine Features Comparison (ERIC document ED 413 900) (1997, PDF)
• University of Waterloo, David R. Cheriton School of Computer Science — Web Search Engines, CS748T survey presentation (2001, PDF)
• en.wikipedia.org/wiki/AltaVista
• theregister.com/2013/06/29/yahoo_to_close_altavista/
• washingtonpost.com/business/technology/farewell-to-altavista-yahoo-kills-old…
• thenextweb.com/news/yahoo-to-shutter-ancient-search-engine-altavista-as-part…
• en.wikipedia.org/wiki/Louis_Monier