Profound's Valuation Rose 1.8x in Seven Months While It Says Revenue Tripled

Seven months, 1.8x the valuation, and revenue the company says tripled

Profound, a US startup that builds marketing software for AI search, announced on Sept. 15 that it had raised a $180 million Series D at a $1.8 billion valuation. Sequoia Capital and Kleiner Perkins co-led the round, and existing investors including Lightspeed Venture Partners, Khosla Ventures and South Park Commons also took part.

The previous round was announced on Feb. 24. Profound raised a $96 million Series C led by Lightspeed at a $1 billion valuation. According to Fortune, that brought total funding to $155 million, and the company had fewer than 120 employees. It counted more than 700 enterprise customers, including 10% of the Fortune 500.

The figures moved in under seven months. According to TechCrunch, Profound says its revenue tripled over the past six months and that it now has more than 1,000 enterprise customers, naming Comcast, The Estée Lauder Companies and Walmart. Over roughly the same stretch, the valuation rose 1.8x. Profound has not disclosed revenue, and the two periods do not line up exactly. Taking the company’s description at face value, though, investors paid less per dollar of revenue than they did in February.

From a dashboard to an agent that does the work

Profound started two years ago as an analytics tool showing how brands appear in AI answers. Its product page says it collects answers by entering prompts into the same front-end experiences ordinary users see, rather than calling APIs. It covers nine engines: ChatGPT, Perplexity, Claude, Microsoft Copilot, Google AI Overviews, Google AI Mode, Google Gemini, Grok and DeepSeek, in more than 30 languages and more than 150 regions.

With the Series C in February, it launched Profound Agents, which turn that data into marketing copy and distribute it. This time the lead product is AI Marketer, or Aim. In a post on the company blog, CEO James Cadwallader wrote that Aim works across a company’s data, notices what is going well and what is not, and brings a plan. Once a marketer approves, it writes the brief and a first draft, routes it to whoever signs off, publishes it to the content management system, and reports a month later on what it moved. The company is holding a live demo on Sept. 30.

The pricing page points the same way. Only two plans are listed. The free trial runs 10 prompts once, on ChatGPT only. The enterprise plan, which tracks up to nine answer engines, is priced per customer. There is no self-serve monthly tier in between.

Half of July’s cited domains were not there in June

The thing being measured also keeps shifting. In an analysis published in 2025, Profound compared the domains AI answers cited for the same prompts in June and in July. Roughly 40 to 60% of the domains cited in July had not appeared in June. Comparing January with July, the figure grew to 70 to 90%. The company concluded that checking only now and then is essentially meaningless.

In a recent blog post, Profound said half of the content answer engines cite is less than 13 weeks old, and it launched a feature showing how long a page’s citations take to fall to half of their peak.

All of these numbers come from a company that sells the measurement tool, and the conclusion that volatility is high leads straight to the case for continuous monitoring. They are safer to read as a warning against setting content strategy on a single measurement than as independently verified figures.

Where the lost clicks show up first

As AI answers take up more of the page, clicks fall. Reach, publisher of the UK’s Mirror and Express, said on Sept. 16 it would cut another 220 editorial jobs. According to The Guardian, Reach said earlier this year that a slump in digital page views was driven by a 46% year-on-year decline in traffic from Google, as Google’s AI Mode and AI Overviews remove the need for readers to click through. Reach said it would move away from page views toward active engaged time as its main measure.

Profound sells the other side of that trend: help keeping a brand’s name inside the answer even as clicks shrink. Cadwallader described the content AI agents want as current, boringly specific and reasoned from first principles. For a sneaker, that means ignoring phrases like “the most stylish” and giving the weight of the shoe, the material, and how it fares for high arches versus low arches.

In Korea, Naver’s AI Briefing sits outside the list

For founders selling into Korea, the list of tracked engines matters. Naver is not among Profound’s nine. Naver introduced AI Briefing, which summarizes search results, in March 2025. According to Yonhap News, AI Briefing was applied to about 20% of Naver’s integrated search queries as of the end of last year and had more than 30 million users. Naver plans to expand it to about 40% of all searches this year. Google also began rolling out AI Mode in Korean on Sept. 8, 2025.

A team selling to Korean consumers cannot tell from a global tool’s share-of-voice chart how Naver’s answers describe its brand. A team targeting the US market should look at ChatGPT and Google first. Either way, if cited sources shift this much within a month, how often you check matters more than any visibility you win once. Writing down the questions customers are likely to ask and running them through each engine on a regular schedule is something a team can start without a paid tool.