Time 开始向 AI Agent 投放广告

Time 已开始向 AI Agent 投放广告。Ally Bank 和 Project Management Institute 是首批购买方。

这听起来很奇怪，但也合乎逻辑。要理解其中原因，重要的是回到事情的起点。

上个月，Time 开始将其所有网页从 HTML 转换为 Markdown 版本——即每个页面去除设计和图片、仅保留文本的精简副本，从而让 AI 系统和 Agent 更容易读取。该出版商的逻辑是，访问越便利，其出现在生成式搜索查询结果中的机会就越大。

不过，这一效果也有一定限度。Markdown 页面究竟能在多大程度上抵御搜索流量下滑的威胁，目前仍无定论。Time 在等待答案的同时，选择在这些页面中加入广告，以从不断增长的机器人流量中获取部分收入。为此，该出版商正与 AI 驱动的广告技术平台 Mobian 合作，将页面转换为 Markdown 并生成 Agent 广告。广告本身采用 FAQ 格式，包含品牌信息和营销信息，并标注为赞助内容。

Time 和 Mobian 称，这是出版商首次投放专门面向 AI Agent 的广告。更多广告已在筹备中。Time 首席运营官 Mark Howard 告诉 Digiday，Time 的销售团队正在向那些已将企业网站转换为 Markdown 文件的品牌推介 Agent 广告，因为这些品牌已经在思考如何触达 AI 机器人。他表示，Agent 广告将成为触达 LLM 的另一种方式。根据 Mobian 联合创始人兼 CEO Jonah Goodhart 的说法，Mobian 的数据显示，大约 15% 的品牌正在运营自己的 Markdown 页面。随着越来越多品牌押注于“双轨互联网”，这一比例很可能会上升。

“影响 Agent 或许比影响人类本身还要重要，因为你影响一个人类时，影响的只是一个人。当你影响 ChatGPT 时，潜在影响的是整个 ChatGPT。如果 ChatGPT 改变了它对某个品牌的表述，影响将非常巨大，超过任何一场单独的营销活动所能做到的程度，”Goodhart 表示。“这里的核心理念是，LLM 和 AI 需要关于品牌的可信信息。我们正以一种非常直接、简便的方式向它们提供这些信息。”

换句话说，这些广告为 Ally 以及非营利专业组织 Project Management Institute 等品牌提供了另一种工具，可用于影响它们在生成式搜索中的呈现方式。LLM 可以从广告中获取并索引品牌信息，继而塑造 AI 生成的回答，最终影响使用这些 AI 问答引擎的人类。

这也是为什么 Time 内部将 Agent 广告视为其更广泛的面向品牌销售 GEO 产品计划的一部分。该产品建立在 Time 对 AI 搜索引擎从何处提取信息、以及 Time 何时出现在这些摘要中的了解之上。这些努力最终能否成功，部分取决于这些广告能否将 Time 所看到的激增机器人流量变现。

Howard 拒绝披露这波流量增长背后的具体数字，但援引 TollBit 的数据称，Time 收到的针对其内容的 AI 爬虫请求量，高于 TollBit 网络中近 7,000 家出版商网站中的大多数。Howard 表示，当 Time 发布 Time100 系列榜单时——例如本月发布的 Time100 Creators——机器人流量会出现特别显著的峰值。事实上，Time 在多数日子里看到的机器人流量已超过人类流量，这与 Cloudflare 的数据相呼应；后者显示，目前超过一半的网络流量已是机器人流量。

“这是一个不断增长的流量来源，因此也是不断增长的广告库存来源。如果我们能够继续以此前的方式开发这一产品，使之达到能够向这些品牌提供产品的程度，我们相信它将拥有巨大价值，”他继续说道。

即使这一模式未能成功，也不会是因为流程不够简单。Mobian 能够根据品牌简报生成 FAQ 格式的 AI 品牌广告。这则 Agent 广告会被转换为 PDF 供人工审核，随后与客户共享并获得批准——Howard 表示，这与标准品牌内容广告的流程并无太大不同。之后，Mobian 会向 AI 搜索引擎提出同样的 FAQ 问题，以衡量广告表现，并持续追踪可见度、好感度和准确性评分。Howard 表示，Agent 广告可以根据 Time 的系列榜单进行语境定向，也可以按日期范围定向。

而且，它们不会便宜。Time 正在为这些 Agent 广告收取溢价。Howard 表示，其逻辑在于，营销人员支付的是塑造 LLM 所检索信息的机会；由于权威内容上的 AI 机器人曝光稀缺且有价值，这种稀缺性为广告主创造了施加影响的空间。Time 每个 Markdown 页面只销售一则 Agent 广告，并衡量该页面的 AI 机器人流量，但 Howard 拒绝透露具体定价细节。

“这是榜单赞助或系列赞助的一个显而易见的延伸，因为我们确实知道 Agent 的到来会带来流量峰值，”Howard 表示。“按语境进行匹配的能力会成为非常有帮助的工具。而域名权威性、流量规模、投放一场营销活动并据此进行定向的能力，所有这些因素正汇聚在一起。我们认为，这将成为我们面向品牌合作伙伴提供的服务中极其重要的一部分。”

其重要性究竟有多大，仍有待观察。鉴于这类做法此前几乎没有先例，目前尚不清楚 LLM 是否会以不同方式对待广告和编辑内容，甚至是否会在决定检索哪些信息时降低推广材料的权重。但对广告出现位置和方式进行严格限制已有先例：Google 在 2010 年代开始对出版商网站上的广告执行更严格的标准和垃圾内容政策。

波士顿咨询公司技术构建与设计部门 BCG X 的董事总经理兼合伙人 Rob Derow 也意识到了这一风险。他表示，Markdown 广告最大的风险在于，尚无规则规定 LLM 应如何处理这类广告，以及这种处理方式未来可能如何改变。如果 LLM 最终将其视为一种伪装（cloaking）形式——即网站向搜索引擎爬虫提供与人类用户所见内容不同内容的 SEO 手法——出版商可能会发现这些页面的效果下降，或因在 Markdown 文件中放置赞助内容而受到处罚。不过，Derow 表示，可以借助 AI 可见度工具轻松监测这一情况。

考虑到这一点，Time 的 Agent 广告在顶部包含“赞助内容”披露，并注明品牌客户名称，尽管目前尚无政策要求必须使用这一标签。Howard 和 Goodhart 指出，当 AI Agent 从出版商网页中检索信息时，广告实际上已经在被投放给它们。

“由于这是一件全新的事情，我们现在还不知道答案；我们相信自己正在开辟第一条道路，”Howard 表示。“但我们正把过去处理这类广告时获得的经验应用到这里。至于最终是否会需要这样做，我们还不知道。”

Time has begun serving ads to AI agents. Ally Bank and Project Management Institute are among the first buyers.

Strange as that sounds, it’s also logical. To understand why, it’s important to go back to the start.

Last month, Time began converting all its webpages from HTML into markdown versions — stripped-down, text-only copies of each page that drop the design and images, making them easier for AI systems and agents to read. The easier that access, went the publisher’s thinking, the better its chances of showing up in generative search queries.

Well, to a point. The jury is still out on how much of a bulwark markdown pages can be against the threat of falling search traffic. Putting ads in those pages is Time’s way of squeezing some revenue out of its growing bot traffic while it waits to find out. To do that, the publisher is working with AI-powered ad tech platform Mobian to convert its pages into markdown and generate the agent ads. The ads themselves are formatted as FAQs with a brand’s information and messaging, and labeled as sponsored content.

Time and Mobian claim this is the first time a publisher has served ads specifically targeting AI agents. More are already being lined up. Mark Howard, Time’s chief operating officer, told Digiday Time’s sales team is pitching agent ads to brands that have already converted their corporate websites into markdown files, as those brands are the ones already thinking about how to reach AI bots. The agent ads would be another way to reach LLMs, he said. Mobian’s data shows about 15% of brands are powering their own markdown pages, according to Jonah Goodhart, co-founder and CEO of Mobian. As more brands bet on a two-track internet, that figure will likely rise.

“Maybe it’s more important to influence the agent than even the human, because with a human you influence one person. When you influence ChatGPT, you’re influencing potentially all of ChatGPT. If ChatGPT changes what it says about [a brand], it’s massive and it’s more than any one campaign could ever do,” Goodhart said. “The big idea here is LLMs and AI want trusted information about brands. We’re serving that up to them in a very straightforward, easy way.”

Put another way, the ads give brands like Ally and the non-profit professional organization Project Management Institute another lever to pull on how they show up in generative search. LLMs can pick up a brand’s information from the ad and index it, which can then shape AI-generated responses, which in turn, can influence humans using those AI answer engines.

It’s why the agent ads are viewed internally as part of Time’s wider push to sell a GEO product to brands, built on what it’s learned about where AI search engines pull information and when Time shows up in those summaries. How successful those efforts turn out to be will depend partly on how well the ads can monetize the surge of bot traffic Time has seen.

Howard declined to share the numbers behind that surge, but cited TollBit data showing the publisher sees more AI crawler requests for its content than the majority of the nearly 7,000 publisher sites in TollBit’s network. Time sees an especially large spike in bot traffic when it launches a Time100 franchise list, such as the Time100 Creators published this month. So much so, in fact, that Time sees more bot traffic than human traffic most days, Howard said, mirroring Cloudflare data that shows more than half of all web traffic is now bot traffic.

“This is a growing traffic source, and therefore a growing source of inventory that we believe has huge value if we can continue to develop this product the way we have to get to this point to be able to have an offering to these brands,” he continued.

If it doesn’t get there it won’t be for lack of a simple process. Mobian can generate an AI brand ad in an FAQ format based on a brand brief. That agent ad is converted into a PDF for humans to review, then shared and approved by the client — not unlike the process for a standard branded content ad, Howard said. Mobian then puts those same FAQ questions to AI search engines to measure how the ad performs, tracking visibility, favorability and accuracy scores over time. The agent ads can be targeted contextually, against Time’s list franchises or by date range, Howard said.

And they won’t be cheap. Time is charging a premium for these agent ads. The thinking is that marketers are paying for the chance to shape the information LLMs retrieve since AI bot impressions on authoritative content are scarce and valuable, Howard said, and that scarcity is what gives advertisers the opening to influence it. Time sells one agent ad per markdown and measures AI bot traffic to that page, though Howard declined to share the pricing details.

“This is an obvious extension of list sponsorship or franchise sponsorship, because we do know that there is a spike of the agents coming,” Howard said. “That ability to contextually align becomes a very helpful tool. And given that domain authority, the volume, the ability to flight a campaign and to target it accordingly, it’s all coming together in a way that we think is going to become a really important component of what our offering to our brand partners is.”

How important remains to be seen. It’s not clear if LLMs treat ads differently from editorial content, or even penalize promotional material when deciding what information to retrieve given it hasn’t really been done before. But there is a precedent for cracking down on where and how ads can appear: Google began enforcing stricter standards and spam policies for ads on publishers’ sites in the 2010s.

It’s a thought not lost on Rob Derow, managing director and partner at BCG X, Boston Consulting Group’s tech build and design unit. He said the biggest risk with markdown ads is that lack of rules governing how LLMs treat them – and how that could change. If LLMs eventually view this as a form of cloaking (an SEO tactic where a website delivers different content to search engine crawlers than what is shown to human users), publishers could see those pages become less effective, or be penalized for putting sponsored content in the markdown files. However, Derow said this can be easily monitored using AI visibility tools.

With that in mind, Time’s agent ads contain a “sponsored content” disclosure at the top naming the brand client, even though there’s no policy yet requiring that label. Ads are already being served to AI agents when they retrieve information from publishers’ webpages, Howard and Goodhart noted.

“We don’t know yet because this is brand new, and we believe that we are paving the first path forward here,” Howard said. “But we are taking learnings from how we approached this kind of advertising in the past and applying it here, and whether or not you are going to end up needing that or not, we don’t know.”