Liner, a South Korean AI search engine, has been testing ads in the U.S. and Europe using a traditional cost-per-click model. But unlike Perplexity, which is working to attract advertisers with a cost-per-mille (CPM) model, Liner thinks CPC is more measurable.
The company is testing three formats: ads that show up above and below an AI-generated answer and another ad format called “generative ads” that shows up within an answer. Tests for ads began last year in the U.S., with generated ad tests since earlier this year.
“We had this kind of hypothesis,” Liner founder and CEO Luke Kim told Digiday. “In digital ads, there are search ads and display ads. Search ads are a good business because the conversion rate is high. Display ads are a good business because it has a lot of impressions and a lot of views because you can put banners anywhere. Maybe we can create some kind of third category, which is AI search ads.”
Liner wouldn’t share the exact number of advertisers, but said it’s tested ads against more than 20 million searches worldwide through various ad formats. When asked about Liner’s partnerships with ad tech companies, agencies or advertisers, Kim wouldn’t disclose any specifics. However ads served in response the a few Digiday queries to Liner showed ads for HBO Max, GrubHub, Nike and an Amazon page for a book titled the “World Atlas of Coffee.”
Instead of retargeting users, Liner serves ads based on just the most recent question a user asks, which Kim said has the most immediate interest and value to both the user and the advertiser. Liner might explore serving ads based on a user’s past search history for more refined targeting. Kim added that using more than just the last query could also provide context that might help if the user’s query is ambiguous.
“We’ve learned the key is understanding the question itself,” Kim said. “People don’t ask questions in really good ways. The question itself is sometimes really bad, but even though we have to give them a great answer.”
Liner has both a free version and a subscription version. As competitors started giving away more premium features for free, Liner decided it should start testing ads as a way to generate revenue while also offering more within its free tier. The alternative was to try making money with the enterprise category.
Last fall, Liner raised $29 million from investors including Samsung Ventures, with total funding around $44 million. But that’s far less than capital raised by its rivals. Liner’s monthly subscription price of $36 is also higher than the $20 charged per month by OpenAI, Perplexity and Anthropic.
“If we look at the long long game, this subscription might not be our primary source of revenue because the competition is going to cut away the margins,” Kim said. “…Our team is more focused on building consumer products and we’re not experts in enterprise. As a startup we wanted to focus on one thing, so we went to ads.”
When it comes to AI-driven search, there’s been plenty of debate. There’s the question whether newer players like Perplexity are taking the right approach by shifting to a CPM model. But there have been questions around whether giants like Google are moving too fast with ads within AI Overviews. Publishers worry about threats to referral traffic. Users are worried about answer accuracy.
When asked about choosing CPC over a CPM model, Kim said CPC’s better for measuring ads and even if CPM is easier to implement as a publisher. He added that Liner’s click-through rate is between 1% and 2%.
CPC is still more comfortable than CPM for most advertisers used to bidding on Google Search, said Brad Jashinsky, director analyst at Gartner. AI search engines should focus on helping advertisers generate and track both conversations and conversion value. However, he also warned marketers shouldn’t get distracted by “shiny object syndrome” — especially since AI search startups only account for a small fraction of ads compared to giants like Google and Bing.
“The key for AI search engines is to drive users to take actions from ads and ultimately deliver higher return on investment than other advertising channels,” Jashinsky said. “The last thing advertisers want are new metrics that are not comparable to other channels.”
Growing pains — and gains
Liner says it now more than 10 million users, with two thirds of those based in the U.S. The company also noted 90% of its user base are students and researchers with college use across the country and schools like the University of Southern California, University of Michigan, UC-Berkeley, New York University and Texas A&M.
Last month, Liner’s website had 6.5 million visits in January, according to SimilarWeb data. That’s still small when compared with competitors. Last month, Perplexity had 13.5 million visits, ChatGPT had 330.2 million and Google had 3.2 billion.
As part of its efforts to scale Liner, the company recently began working with what Kim only described as “one of the largest manufacturers and telcos around the world.” However, he wouldn’t disclose whether it was Samsung or another company. Liner also has targeted college students that might be interested in using the platform as early adopters, especially those interested in using Liner for research.
When it comes to the overall search landscape, keyword-driven intent is still the highest converting and highest value traffic for most industries, according to Jim Yu, CEO of the SEO platform BrightEdge. If someone is searching for the best 50″ TV in my living room, Yu said it’s because they’re looking to buy a TV. On the other hand, he noted ads within an answer might help create brand narrative and also provide ways to foster follow-up questions: “This isn’t yesterday’s CPM model.”
Liner might be able to fill a gap by answering questions quickly, Yu said. He also noted that Liner’s connection with Samsung might help drive mobile phone usage, which then could help since a lot of mobile searches are related to transactions and navigation. Location-based answers will also help.
“People aren’t willing to wait that long for a lot of things when they’re sitting down on their desktop,” Yu said. “On mobile, it’s much more latency-sensitive and location driven.”
Agencies that Digiday spoke with hadn’t tried Liner yet and hadn’t bought ads for clients. Though media buyers said they see the sense in taking a CPC-based, intent-based approach. They also wonder if search will become brand-driven or stay focused on performance. Another challenge is whether testing ads for new search engines will be worth it, especially if there’s a lot of work in setting up everything from setting up accounts and developing creative to tracking performance.
“From an agency POV, the intent-based model has worked for a long time and makes a ton of sense,” said Sam Huston III, svp creative & media at DEPT. “… But it’s going to be hard to get people to pay for search.”
Others say clients aren’t yet interested in testing AI search or new AI platforms until there’s more proof that they work. They also say advertisers are worried about running ads within generative search unless they can make sure AI-generated answers won’t paint brands negatively.
“We don’t need to be the first to test,” said Jeff Eisenfeld, head of media at Mother. “But if it’s something that’s been successful and then we see that within our competitors or within our verticals and that makes sense, we’re not going to be scared to test.”
https://digiday.com/?p=570227
