Yahoo claps back on AI search engines with Yahoo Scout

Yahoo may not be the most headlined company in tech anymore, but its reach can’t be denied. With nearly 250 million monthly users across the country and 700 million globally, it’s still the second most popular email client in the world, and the third most popular search engine in the U.S. (even though that search engine has technically been powered by either Bing or Google since 2009).

As a privately owned company since 2021 (once worth $125 billion, but purchased for a mere $5 billion at the time), its CEO Jim Lanzone says that the last few years have been about “getting the house in order.” But now, he promises, “this is one of the biggest turnarounds people have tried in internet history.”

Lanzone says that turnaround begins with Yahoo Scout, which launches today in beta. In short, Yahoo Scout is a new, free AI search engine (it’s also an omnipresent button across Yahoo verticals like Finance, Sports, and Mail) that’s there to summarize the performance of a business or break down the key moments of a game.

In one mode, it’s essentially Yahoo’s version of Claude or ChatGPT. (Yahoo Answers for the AI era!) In the other, it’s an AI-translate button accompanying Yahoo’s editorial content, boiling down articles into takeaways. It will even summarize the sentiment of the comments on stories you read across Yahoo.

“We [aren’t] the first to market here, but in evaluating whether we should keep outsourcing or build the AI layer ourselves, it just became clear that we could do this best for our users,” says Lanzone. “We had a lot of unique assets to do that. And so in that context, timing is almost irrelevant, right? Because this is about Yahoo users on Yahoo, searching on Yahoo, versus what they were getting before.”

Led by Eric Feng, SVP & general manager of the Yahoo Research Group—who is best known for recruiting and leading the technical team that created Hulu—it’s powered by a combination of Yahoo’s knowledge graph, Anthropic’s Claude, and Bing’s open web APIs. (Yahoo says that user data is kept internally and does not train Claude or Bing.)

Yes, that means Yahoo is still relying on external partners, but the team says that search will feel like Yahoo because it’s so based within the Yahoo ecosystem. Specifically, Yahoo Scout can answer a question with everything Claude knows within its LLM (and gosh, does Yahoo Scout love to build a table breaking down information!) And it can also search the web itself as well as Bing. 

But the team insists that Yahoo’s own knowledge graph provides a lot more on top of Claude and Bing today, and it promises even more into the future. Specifically, Yahoo publishes 30,000 pieces of licensed stories and other content each day that are at its fingertips, and users create 18 trillion “events” across its services each year. “Every event you get just makes the overall totality of the experience smarter,” says Feng.

Even if it’s something as simple as searching for the score of your favorite team, that’s actionable data for Yahoo in the short term, but even more so in the medium to long term as it shifts Scout from a generalized AI search engine or summarizer to more of a personal AI.

“Just to be direct about it, you will see the roadmap include personalization [of AI],” says Lanzone. “That’s certainly where the category is headed, and it is a unique asset to Yahoo to be able to already have that built in. You know, we’re not trying to acquire an audience to start to build that up from scratch. We actually have all that knowledge and that relationship to dip into then layer this on top of.”

The experience of Yahoo Scout, and what that means for publishers

Yahoo Scout lives as a responsive website and also as a standalone app.

As a search engine, Scout’s white search bar is accompanied by an ever-changing rotation of animated clipart (yeehaw, it’s quite a cute cowboy hat), bringing back some of the brand’s original quirk.

The rest of the experience will be pretty familiar to anyone who has used an LLM, as every query is less a collection of blue links than an explanation of an idea.

From a demo I watched, it really does have a bias to break searches down into comparative tables (which may be useful or cloying, depending how often that approach is taken). But it’s also less text heavy than many LLMs, because it makes liberal use of thumbnails listing news stories and products to buy. 

That penchant for prominent, colorful linking isn’t coincidence. Publishers are amidst a 30-year mass extinction event, spurred on most recently by AI search tools that mine their reporting and present it to a public that no longer needs to click through to the actual story. Lanzone argues that prominent linking is key to protecting the health of the publishing industry—publishers that Yahoo relies upon for its core media aggregation business.

“Our No. 1 job is to bridge the gap for publishers . . . [with] a beautiful UI that’s very rich and intuitive but still leans into linking out without cluttering the page,” he says. “We’ll see if this is the exact version of where we wind up, but from day one, that’s a priority for us. And then [priority] two is, can we bring search advertising along as well?”

Speaking from two decades in digital publishing, I know that what Yahoo has shown thus far is certainly not enough to protect publishers—and I challenged Lanzone on this point. Yahoo’s design may improve clickthrough rates a hair, but not nearly enough to support the publishing industry. Cloudflare research on search engine referrals demonstrates just how dire this moment is: For every 70,900 times Anthropic bots scraped information off a website, users clicked through only once. Much like McDonald’s couldn’t survive if Uber Eats began distributing its hamburgers for free, reported news can’t live on if search engines are snatching their insights without meaningful compensation.

“It’s not in version one, but we are also passionate about the idea that search advertising was really effective for both publishers, search engines, and advertisers. And there should be a way to cross the chasm, and to bring that model along with generative AI search engines” says Lanzone. “I don’t know. It won’t wind up exactly where it was, but it can be a lot closer to it than we’ve seen to date. And we are actually working on versions of that. It’s just going to be very hard to scale nine-step agentic processes where you get paid downstream, and it’s just a very difficult model to really scale up.”

And make no mistake, Yahoo is advertising with Yahoo Scout. While Anthropic has yet to introduce ads and OpenAI is only starting to, Yahoo is including ads in Scout searches from day one. Those ads appear in traditional paid links in the feed (think Google Adwords). It’s also monetizing its shopping referrals (thanks to an AI shopping platform Vetted it acquired late last year). These are just the beginning of Yahoo’s extensive advertising roadmap, which will be realized with more intricacy later this year.

Outside search, though, Yahoo believes that Scout living across Yahoo services will offer indirect benefit to monetization. “If you are on Mail, Finance, or Sports, we want the technology that we’re providing to make that experience better, and then we get more engagement,” says Feng. “We get those users sticking around longer, and then those properties are able to better monetize.”

For now, Yahoo believes that monetization will be significant enough that it justifies keeping Scout free. That said, even Yahoo cannot resist the allure of subscription service revenue, and suggests that a premium paid version of Scout could live somewhere in the future.

“Look, we have subscription products of finance and sports and places, and you can definitely see a version of that coming here,” says Lanzone. “But 100% of our effort [for launch] has been the evolution of Yahoo search into Yahoo Scout. And we have the luxury of doing that because we have a really good business search already.”

View the full article