A $1.8B Valuation Says AI Search Is Now A Real Channel

On Tuesday, September 15, a two year old company called Profound announced a $180 million Series D at a $1.8 billion valuation. Sequoia and Kleiner Perkins led it. Lightspeed, Khosla and South Park Commons came along.

The part worth sitting with is the timing. Profound had raised a $96 million Series C less than seven months earlier. The valuation roughly doubled in the time it takes most seed companies to close one round.

So what does Profound sell? Software that helps brands show up inside AI search results. Not Google rankings. The answers that ChatGPT, Gemini, Claude and Perplexity hand back to people who asked a question.

The company says revenue tripled in six months and it now has more than 1,000 enterprise customers, with Comcast, Walmart and Estee Lauder on the list.

Comcast is paying for this. Walmart is paying for this. That is the signal, not the valuation.

Because the money is telling you where buyers now start. Enterprises do not fund a new budget line unless something they relied on stopped working.

Here is the short answer. The page that used to rank is now a sentence inside somebody else's answer. If you are not in that sentence, you were never in the consideration set, and you will not see it in your analytics because there was no click to see.

Profound's round is not interesting because a marketing tool got expensive. It is interesting because a thousand large companies concluded they have a visibility problem in a channel that did not exist three years ago.

You have that problem too. You just have less budget and, oddly, a better shot at fixing it. Small companies get cited by AI assistants more easily than big ones, for reasons we will get to.

Less than the hype suggests, more than you would guess, and growing faster than anything else in the mix.

Similarweb's 2026 tracking puts AI referral traffic up roughly 527% year over year. That is a real number off a small base, which is exactly what an emerging channel looks like right before it stops being emerging.

The bigger story is what is not happening. Zero click searches hit record levels, with well over half of US searches in 2025 ending without a click to any outside site. Google searches that show an AI Overview push zero click rates to around 43%. Inside Google's AI Mode, the reported figure is close to 93%.

Read that again. Nine out of ten queries in AI Mode end with the person having their answer and no reason to visit anyone.

Your future traffic chart will not look like your old one. Plan for that now.

The visits that do come through appear to be unusually good ones. Similarweb's comparison of US transactional sites found ChatGPT referrals converting at about 7% against Google's 5%.

Session behavior tells the same story. Visitors arriving from ChatGPT spent around 15 minutes on site versus 8 from Google, and viewed roughly 12 pages versus 9.

That tracks when you think about the path. Somebody asked an assistant a real question, got an answer, and then chose to click through to one named company out of everything the model could have said. They arrive already half sold.

One caution. The mix is moving fast. ChatGPT's share of measurable B2B AI referrals averaged about 63% across March and April 2026, with Claude near 18% and Gemini near 11%, according to SearchSignal's benchmark. A year earlier ChatGPT was dominant enough that nobody bothered measuring the rest.

Do not build a strategy around one assistant.

Profound's customers are buying dashboards. You cannot afford a dashboard and you do not need one yet. What you need is to know what the models say about your category today.

Sit down for an hour and do this by hand:

1. Write the ten questions a buyer would ask before they know your company exists. Not "what is the best CRM." More like "how do dental practices handle recall reminders without a receptionist."
2. Ask each one in ChatGPT, Gemini, Claude and Perplexity. Fresh sessions, memory off.
3. Write down every company named, in order, for each model.
4. Note which sources the assistants cited. Those are the pages shaping what gets said about your market.
5. Repeat the whole thing in thirty days.

That last step matters more than the first four. One snapshot is trivia. Two snapshots is a trend line.

Keep the results in a spreadsheet, a Notion table, or a planning tool like Foundra that walks first time founders through positioning and go to market one section at a time. The tool is not the point. Doing it twice is the point.

Specificity, mostly. Models reach for the clearest match to a narrow question, and narrow is where small companies live.

A few things that seem to move the needle.

Be describable in one sentence containing a category and a customer. "Scheduling software for mobile dog groomers" gets retrieved. "The operating system for modern service businesses" does not, because nobody asks for that.

Get named in places the models already read. Comparison posts, industry roundups, Reddit threads, review sites, podcast show notes. A mention on a page an assistant already trusts beats a page on your own site that nobody links to.

Publish the boring answer pages. Pricing that is actually a number. A real integrations list. Who the product is not for. Assistants cite pages that resolve a question cleanly.

And keep your facts consistent everywhere. If three sources give three different founding years and two different headquarters, models hedge and name somebody else.

Stop writing 2,000 word posts that answer the question in paragraph nine. Assistants extract answers. Bury yours and it does not get extracted.

Stop chasing broad head terms. "Best project management software" is a knife fight between companies with nine figure budgets. You will not win it, and it was never going to convert anyway.

Stop treating your own analytics as ground truth. A large and growing share of the people who know your name never visit your site before they contact you. If your only measurement is sessions, you are managing a channel you cannot see.

Ask new customers one question on the intake form: how did you first hear about us? Free text, not a dropdown. When "ChatGPT suggested you" starts showing up, you have your data.

And stop waiting for this to settle. It will not settle soon, and the companies getting cited in 2027 are building that reputation right now.

Three things, in order of importance to you.

First, a distribution channel opened and it is still cheap. Enterprise budgets are arriving, which means it will stop being cheap. The window where a well positioned small company can get named alongside incumbents is open now and closing.

Second, the skill being paid for is not technical. Profound sells measurement and strategy, not a plugin. The underlying work is knowing what your category's buyers ask, then making sure the accurate answer includes you.

Third, valuation moves like this one tend to mark the middle of a shift rather than the start. Seven months between rounds at nearly double the price is what it looks like when investors think they are late.

You are not late. Most first time founders have not run the ten question exercise even once.

• Profound raised $180M at a $1.8B valuation on September 15, 2026, under seven months after a $96M Series C, with more than 1,000 enterprise customers.
• AI referral traffic is up roughly 527% year over year, while zero click rates sit near 43% on Google searches with AI Overviews and close to 93% inside AI Mode.
• The visits that do arrive convert better: about 7% from ChatGPT versus 5% from Google, with longer sessions and more pages viewed.
• The channel is fragmenting, with ChatGPT near 63% of B2B AI referrals and Claude and Gemini taking real share.
• Your free version of what Profound sells: ten buyer questions, four assistants, written down, repeated monthly.
• Specific positioning and third party mentions matter more than the volume of your own blog.

Is AEO just SEO with a new name? There is overlap, since both reward clear pages and credible third party mentions. The difference is the goal. SEO tries to win a click from a results page. AEO tries to get your company named inside an answer where there may be no click at all. Measurement and content shape both change as a result.

Do I need to pay for an AEO tool at seed stage? No. Tools like Profound exist because enterprises track thousands of prompts across dozens of brands. You have one product and maybe ten questions that matter. Do it manually until the manual version takes more than two hours a month.

Should I block AI crawlers to protect my content? That is a real tradeoff and it depends on your business. If your content is the product, blocking may protect revenue. If your content is marketing, blocking removes you from the channel where buyers now start. Most early stage companies are in the second group.

How long before this shows up in revenue? Slower than paid, faster than traditional SEO. Getting mentioned in sources the models already read can shift what assistants say within weeks, but attribution is murky, so measure it through customer intake answers rather than your analytics dashboard.

What if my category is too new for anyone to ask about it? Then buyers are asking about the problem instead of the category. Write your ten questions in the language of the pain, not the solution. That is usually where a new company gets found anyway.

1. TechCrunch: AEO startup Profound hits unicorn valuation, raises $180M Series D
2. Similarweb: AI Search Stats 2026, market share, referral and citation data
3. Similarweb: Zero-Click Marketing, what the 2026 data means
4. SearchSignal: 2026 AI Search Referrals and Citations Benchmark
5. Profound