Before Google became synonymous with web search, there was a wild, experimental period where dozens of search engines competed for dominance. Between 1994 and 1998, the way people found information online was chaotic, fascinating, and very different from what we know today.

WebCrawler launched in 1994 as the first search engine to index entire web pages rather than just titles. It was a revelation at the time. You could actually search for words that appeared anywhere on a page, not just in the URL or heading. The database was small by today’s standards, maybe a few hundred thousand pages, but it felt like magic.

Then came Lycos later that same year, with a spider that crawled faster and indexed more pages. The name came from the wolf spider family Lycosidae, known for hunting prey rather than building webs. That metaphor actually worked. Lycos went out and hunted down web pages aggressively.

AltaVista changed everything when it launched in December 1995. It was genuinely fast, with a massive index powered by Digital Equipment Corporation’s powerful Alpha servers. AltaVista introduced advanced search operators that power users loved. You could exclude terms, search for exact phrases, and filter by domain. It felt professional and serious.

Excite and Infoseek competed by trying to understand search intent rather than just matching keywords. They weren’t very good at it, but the ambition was there. Both companies went public and were briefly worth hundreds of millions of dollars before Google wiped them out.

What’s interesting looking back is how these early search engines monetized. Most relied on banner ads and sponsorships. The idea of paid placement in search results didn’t really take off until GoTo.com (later Overture) introduced it in 1998, creating the business model Google would later perfect with AdWords.

The search results themselves were inconsistent and easy to manipulate. Webmasters quickly figured out they could stuff keywords into meta tags and hidden text to rank higher. Every search engine had different ranking algorithms, so optimizing for all of them was basically impossible. People maintained lists of their favorite search engines and would try three or four before finding what they needed.

Meta-search engines like MetaCrawler and Dogpile emerged to solve this problem by querying multiple search engines and combining results. It was a clever workaround but ultimately a sign that the whole ecosystem was fragmented and messy.

AltaVista remained the gold standard until Google arrived in 1998. Google’s PageRank algorithm, which ranked pages based on incoming links rather than just keyword density, produced noticeably better results from day one. Within a few years, AltaVista was a nostalgic memory and Google was a verb.

What killed the early search engines wasn’t just better technology. It was also the portal strategy. Yahoo, Excite, and Lycos all tried to become “sticky” destinations where users would hang out, check email, and read news. They de-emphasized search in favor of becoming portals. Google did the opposite, focusing obsessively on search quality and keeping its homepage minimal and fast.

There’s something lost from that early era. Search engines today are incredibly powerful but also monolithic. Google handles over 90% of searches globally. In the late 90s, there was genuine competition and experimentation. Search results varied wildly between engines, which was frustrating but also meant no single company controlled access to information.

The old search engines are mostly gone now. AltaVista was acquired by Yahoo and eventually shut down in 2013. Lycos still exists as a zombie brand. Excite got absorbed into various corporate entities. WebCrawler somehow survives as a skin over other search engines.

If you want to experience what early search felt like, the Internet Archive has captures of these sites from the 90s. The results are frozen in time, but you can see the interfaces and get a sense of how different the web was when finding information required more luck and persistence than algorithmic precision.