The arc from 1997 launch to 2001 shutdown offers a cautionary parallel for firms that bolt generative-AI features onto existing portals: distribution can boost traffic, but only a distinctive engine keeps users loyal.
At a Glance
- CNET launched Snap.com in 1997 as a full-service portal built on licensed search technology.
- NBC took a 19 percent stake in 1998 and later folded Snap into NBC Internet after committing US$380 million in on-air promotion.
- Snap/NBCi never owned a proprietary crawler, relying instead on an Inktomi licence while rivals refined in-house search.
- Online advertising collapsed in 2000-01, losses widened, and NBC took NBCi private in 2001 at US$2.19 a share.
- The episode underscores that distribution and marketing cannot replace owning the core engine—whether search in 2000 or AI models in 2025.
Origins: CNET’s Portal Gamble
CNET entered the consumer-portal race in 1997, unveiling Snap Online as a single destination for search, news and shopping, according to Wired. Management framed the site as a challenger to America Online and Yahoo, aiming to seed distribution deals rather than pay for mass advertising.
Product managers assembled features quickly by licensing components. The web index came from a third-party crawler, while editors maintained a directory of recommended links. A bundled email service, discussion boards and a nascent shopping hub rounded out the pitch.
CNET’s approach mirrored other portals: familiar layout, customizable start page and hand-curated headlines. What Snap lacked was a proprietary ranking algorithm that could surface more relevant pages than rivals.
“Our goal is not to say we’re the next Yahoo… We want to build a complete package. We’re going after AOL with a business model that creates lots and lots of distribution points.”
That ambition from co-founder Halsey Minor signalled scale over depth, but depth soon mattered more.More Technology Articles
NBC Buys In
NBC purchased 19 percent of Snap in 1998, gaining option rights to increase the stake to 60 percent. The network presented the move as a natural extension of its audience online, Wired reported.
“This is not a situation of us trying to control content. We see this as a business opportunity.”
Despite that reassurance from NBC chief Bob Wright, analysts focused on Snap’s technology gap. Weeks after the deal, the portal switched to an Inktomi licence rather than fund an in-house crawler, a decision flagged by SearchEngineWatch.“Snap was an airliner on fire, looking to come in for a belly landing.”
Zona Research analyst Vernon Keenan’s assessment captured outside skepticism: broadcast promotion solved capital needs but not product differentiation.From Snap to NBCi
Dot-com enthusiasm peaked in 1999 when NBC rolled Snap, Xoom.com and other assets into a new company, NBC Internet (NBCi). The entity agreed to buy US$380 million in television spots from NBC over four years, as detailed by Wired.
The roll-up produced an early super-portal: search, auctions, personal pages and webmail. Traffic rose each time NBC ran promotional slots during prime-time programming, helping NBCi secure an initial market value beyond US$2 billion.
Management argued that a single sign-on across services would increase stickiness, but users still relied on licensed Inktomi results to find pages. When competitors returned identical links, differentiation reverted to branding.
In 2000 NBCi folded the Snap and Xoom domains into one umbrella site and promised a “personal agent on the web,” according to The Guardian. Consolidation simplified the homepage yet did not address the absence of proprietary search.
Crash-Landing
Online advertising cooled sharply in late 2000. NBCi, dependent on media spend and carrying heavy operating costs, reported widening quarterly losses. Investor sentiment shifted as the dot-com bubble deflated.
NBC moved to limit the damage in 2001, offering US$2.19 a share to take NBCi private and announcing layoffs. Chief executive Will Lansing told reporters that continued public trading would erode value further, as covered by Wired.
“Rather than continuing to operate at a significant loss, and having the value of NBCi continue to erode, we believe this transaction is in the best interest of NBCi’s public stockholders.”
NBC absorbed the portal’s remaining services into NBC.com and retired the Snap and NBCi brands. By year-end, a venture once billed as television’s gateway to the web had vanished from consumer rankings.Strategic Post-Mortem: Snap, Yahoo and Google
Yahoo began as a manually edited directory in 1994 and locked in distribution through early browser deals, an archival timeline at Web Design Museum shows. The directory model evolved into a full portal, but Yahoo retained user loyalty through email and finance products that competitors could not easily replicate.
Google, launched in 1998, focused on speed and relevance by developing the PageRank algorithm, which evaluated the authority of pages by inbound links. Research synthesized in Search User Interfaces attributes the company’s rise to owning that engine rather than layering services around it.
Snap pursued a middle path: heavy media spend without proprietary search. Once Google delivered clearly better results, users defected because Snap offered neither a superior crawler nor a sticky adjacent product.
Yahoo’s later decline illustrates that distribution alone is not permanent, yet its decade-long run underscored how proprietary infrastructure—early directory data and later email—can delay user flight. Snap, relying on contracted search from day one, lacked that buffer.
Lessons for Today’s AI Race
Generative-AI services now appear across news portals, retail sites and productivity suites. Many rely on third-party models accessed through application-programming interfaces rather than in-house training. Snap’s history suggests that such dependence leaves little room to differentiate once multiple firms tap the same engine.
Distribution can still amplify a compelling product. NBC’s broadcasts drove early spikes in Snap traffic, just as social integrations boost usage for today’s chatbots. Yet performance parity undercuts branding once user comparisons become effortless.
Technical sovereignty also shapes cost curves. NBC poured hundreds of millions of dollars into advertising but still paid licensing fees to Inktomi. Firms layering AI widgets on rented models now face similar dual outlays—marketing spend plus per-call usage charges—without compounding returns from owning the algorithm.
Finally, consolidation often follows technical dependence. NBC merged multiple sites into NBCi to mask fragmentation, but the move arrived after consumer attention had already shifted. Companies knitting AI features into legacy portals may confront the same timing risk if they wait for traffic softness before streamlining.
Conclusion
Snap.com’s four-year rise and fall show that control of the engine—not the loudness of the megaphone—anchors durable platforms. Search quality decided the early-2000s internet hierarchy, just as model performance will sort today’s AI contenders. Companies that lean on brand reach while outsourcing the core algorithm risk repeating an expensive lesson NBC learned two decades ago.
Sources
- Chip Bayers. "CNET Takes on AOL, Yahoo with Snap Online." Wired, 1997.
- Jennifer Sullivan. "NBC Buys Into CNET, Snap." Wired, 1998.
- Danny Sullivan. "The Search Engine Report, No. 19." SearchEngineWatch, 1998.
- Reuters. "NBC Rolls Sites into One." Wired, 1999.
- Amy Vickers. "NBC Internet Consolidates Websites." The Guardian, 2000.
- Wired News Report. "Even Cosby Couldn't Save This." Wired, 2001.
- Web Design Museum Editors. "Yahoo! Website, 1994." Web Design Museum, 1994.
- Marti Hearst. "Search User Interfaces, Chapter 1." Cambridge University Press, 2009.
