聊天机器人广告：什么有效，什么无效

Anthropic is holding firm on the pledge made in its Super Bowl commercial this year to keep Claude ad-free. The company’s recent IPO filing didn’t even mention the possibility of selling ads to make money from the AI chatbot. It’s easy to maintain the high ground with about $112 billion in investor cash.

The chatbot industry, though, may find selling ads necessary to defray the high costs of generating AI responses. OpenAI is all in, with over 100,000 customers advertising in ChatGPT since ads debuted there in February. Last week, Google expanded ads in its search chatbot AI Mode to 10 more countries beyond the US and declared that users “find them highly relevant.” Advertisers are far from satisfied with the experience, but they expect it to improve.

Ads stand to be particularly vital for the long-term viability of niche chatbot platforms, which can’t rely on the same degree of venture capital largesse as the top AI labs or the massive revenues of tech giants. I recently spoke with Koah Labs, one of a number of new ad technology companies focused on AI experiences. The San Francisco startup markets itself as bringing ads to over 200 apps, such as DeepAI, which offers tools for creative queries, and Liner, which focuses on providing citations in AI answers. It works with brands like telehealth giant Hims and flower delivery service UrbanStems.

Koah has attracted just a tenth of the advertisers OpenAI has, but CEO Nic Baird still believes he’s getting a good sense of what’s working and what’s fuzzy as chatbots take over from traditional search engines—and what their rise will mean for your AI experiences.

Machine-Crafted

Ads composed partly by AI rather than fully by a human generate double the clicks, Baird says. It’s difficult to assess exactly why, but AI-generated marketing copy likely feels more natural next to the output of an AI model.

Because ads shown by Koah are grounded in product details and imagery supplied by the advertiser, users don’t seem to be encountering AI-hallucinated, erroneous information. But some advertisers, such as drugmakers, face government requirements on what they must—or cannot—say in ads, Baird says. These companies default to more human control over their ads, generating weaker results, according to Baird.

Since ads with dull messages or that feel like clutter could drive users away, the onus is on ad tech companies to find a middle ground that’s engaging to users without triggering risk-averse brands. That balance might be harder to strike when a human is out of the loop.

Creative Targeting

Google searches for “help with stress” are likely to bring up ads for therapists. Chatbots can get more creative, in Baird’s view. For instance, ads for spas have performed well within Koah’s network when shown after users’ prompts related to “stress,” he says.

The key difference, according to Baird, is that Koah can select and customize the ad based on “not only the prompt, or the Google search, but also the response from the AI platform, which brings context from the rest of the conversation thus far.” In addition, chatbot prompts tend to be at least double the length of Google queries, giving Koah more information to determine a relevant ad.

Baird claims users are accustomed to navigating through multiple topics within a chatbot conversation, which could make them more receptive to adjacent marketing than they are in traditional search. Koah says users are clicking on about 3 percent of ads, about on par with the performance of the more mature Google ads system.

All the Context

When called on to show an ad, Koah generally receives a user’s entire conversation up to that point to help inform what it displays. While traditional search queries generally look for specific information, people often use chatbots as not just conversation partners but confidants. It’s unclear whether users will accept the privacy trade-offs that come with personalized chatbot ads as they begin to proliferate.

Baird says Koah anonymizes the queries it receives and stores them in a de-identified way, but he declined to comment any further on the data collection and storage. While acknowledging that personalization can become intrusive, he argues that there are benefits for users and advertisers alike as the tech matures. Nonsensitive information about chatbot discussions, for instance, could inform personalized content on the page that loads after an ad click. Perhaps someone expressed interest in a specific feature, or the AI model didn’t highlight a particular downside of a competing product. The advertiser could fill in the gap.

Koah doesn’t have data on whether people actually want these customized pages, but the tech industry has long argued that people prefer when ad experiences feel relevant. Having the conversation history also could benefit advertisers when Koah enables retargeting, or repeating, an ad to the same user.

Negative Triggers

By Koah’s estimate, one out of every four chatbot queries about a retailer or brand concerns a return, exchange, or other support issue. Baird believes chatbots provide an opportunity not replicated easily elsewhere for advertisers to address what they might politely call a “customer success” matter.

He expects manufacturers and retailers will eventually target disappointed customers to remind them about warranties, engage them in an AI-intermediated support, and potentially steer them away from a more costly customer service interaction. It’s possible that the AI chatbot’s response may provide users with similar help, but buying an ad guarantees companies get their message across.

The ideal outcome is users reaching resolution they can live with more swiftly than through other means. In the worst case, it's a wasted ad buy and a more irate consumer.

Great Unknown

None of the apps Koah currently serves ads in offer users AI agents that can shop on their own. However, apps like Instinct and Meta’s Muse are starting to take over shopping chores for some users. Meta CEO Mark Zuckerberg recently said Muse could become the company’s biggest source of revenue but suggested it would come from transaction fees rather than ads.

That doesn’t mean advertising will be entirely cut out from bot-dominated ecommerce, with initial studies suggesting that ads could influence AI agents.

Baird expects an opportunity to give “the user's agent the best possible recommendation, and that’s going to be a combination of a great discount, a really relevant product,” and an easy digital path to make the purchase. The chatbot must make “abundantly clear” to the agent what content is sponsored or else risk losing user trust, Baird says.

It’s also possible the chatbot ad market fizzles out before agents take hold if more AI developers adopt Anthropic’s stance. Kevin Baragona, CEO of DeepAI, says while ad revenue has been meaningful, he’s not committed to it because the experience just doesn’t feel right yet in terms of privacy, design, and more.

“I don’t think the ads are working terribly well for anyone,” he says.

This is an edition of the Model Behavior newsletter. Read previous newsletters here.