Neeva
» see neeva.com as it lived, on the wayback machine
~ the obituary ~
i knew of neeva as the ex-google guy's paid-search bet. i did not sign up. the product was US-only for most of its life, the $4.95 monthly price sat in dollars, and paying for a search engine in ankara in 2021 was not a category anyone i knew was buying. the credit-card flow alone would have needed a us-billing address in practice. i tracked the launch as a category story and then forgot about neeva for two years.
sridhar ramaswamy and vivek raghunathan founded neeva in 2019 in palo alto. ramaswamy had run google's entire ads business as SVP through the mid-2010s, raghunathan had been VP for youtube monetisation. they raised around $77.5 million over the life of the company, with sequoia and greylock leading the two main rounds. the pitch was that the ad-driven search model had structural incentives to serve the advertiser first and the user second, and a subscription model would let a search engine actually optimise for what the user was looking for.
neeva went into public beta in the US in june '21. the free tier was capped at a small number of monthly searches, the paid tier at $4.95 removed the cap and included a small suite of privacy tools. by early '23 neeva had folded a generative-ai answer layer on top of the search results, which arguably beat google to shipping a mainstream AI search product. that was january '23. the shutdown announcement came in may.
~ the case file ~
| Born | 2019, palo alto |
|---|---|
| Peak | generative-ai answers shipped january 2023, four months before shutdown |
| Killed | 2 June 2023, consumer search turned off |
| Lifespan | 4 years, of which 2 as a public product |
| Made by | Sridhar Ramaswamy and Vivek Raghunathan, both former Google SVPs |
| Killed by | not enough people willing to pay $4.95 a month for search |
~ the paid-search bet ~
the business problem was that convincing anyone to switch search engines is hard in absolute terms and much harder when there is a monthly charge attached. the neeva team knew the number they needed to hit, the acquisition curves they ran did not converge on it. ramaswamy's blog post at shutdown said outright that "throughout this journey, we've discovered that it is one thing to build a search engine, and an entirely different thing to convince regular users of the need to switch to a better choice."
the specific number of paying subscribers neeva ever had is not on the public record. the total user counts leaked in press coverage were in the hundreds of thousands range for beta traffic. the paid-conversion rate was too low to sustain a search-index-scale infrastructure bill. that was the whole story compressed into an accounting line.
~ the snowflake exit ~
snowflake announced the neeva acquisition on 24 may '23 for $185.4 million in cash. by then the consumer product had a public expiry date. the deal was framed publicly as a search-and-AI-for-enterprise-data play, snowflake wanted the team to build natural-language querying on top of customer data warehouses. sridhar ramaswamy joined snowflake as SVP for AI, took over as ceo of snowflake in february 2024.
neeva's consumer search index went dark on 2 june '23, subscribers got refunds on the unused portion of their subscriptions. the browser extension and mobile apps were pulled from the stores around the same window. the neeva.com domain redirected to a snowflake landing page for a while, then eventually to nothing at all.
~ the inheritance ~
in fairness, neeva's shutdown was a natural experiment in whether the ad-driven search model actually was the problem people said it was. the answer on the market side was that maybe it was the problem, and maybe users still would not pay to fix it. cuil in '08 had bounced off the same wall from the free-search direction, blekko from the algorithmic-transparency angle in '14. neeva just added the price tag and hit the same ceiling.
in those days the AI-search category was about to be dominated by chatgpt-plus-bing and perplexity, both launched or scaled inside months of neeva's shutdown. the generative-answer product neeva had shipped in january '23 became, within eighteen months, table stakes across every major search product. neeva was on the correct side of the AI shift and still could not survive the subscription-conversion problem. the search-index-plus-paid-tier bet remains an open engineering question that no one has closed.
~ leave a rose ~
an anonymous tribute. one rose per visitor per day.
@}-->-- ... roses left here
~ leave a wreath ~
a rose fades by morning. a wreath stays on this grave for good, with your name on it.
from $1, paid once through stripe. it keeps the graveyard running: the hosting and the nightly digging. adopt it and the headstone reads "maintained in memory by" you. the groundskeeper plants new wreaths at the next tending, usually within a day.
» one email a week: new graves and the anniversaries of the dead. subscribe
other search: Deja News, Gigablast, Yandex.Slovari, Facebook Graph Search
also died in the 2020s: Gemini Earn, OpenBitTorrent, Project Loon
~ leave a note ~
visitors before you have left these graveside notes. anonymous welcome.