Why Google rankings don't matter as much anymore

A friend of mine runs a DTC supplement brand. They've been doing about $2M in annual revenue, almost all of it from organic Google traffic. Their SEO is meticulous. Claude Hopkins-level attention to keyword research, content clusters, backlink quality. They've got it dialed in.

Last month he called me and said revenue was down 18% year over year. Traffic was holding steady. Rankings hadn't moved. Conversion rate on Google traffic was the same. But fewer people were buying.

We dug into their analytics together and found something I've been seeing everywhere: their direct traffic and their email signups had both dropped sharply. People were still finding them on Google, reading their content, and then... nothing. No repeat visits. No purchases.

Here's what I think is happening and what the early data shows.

A 2025 survey by a research firm (I won't name them, but it's the one everyone in SEO cites) found that 41% of people under 35 start product research on an AI tool rather than a traditional search engine. That's not "some people." That's nearly half.

These people never see your Google rankings. They never see your carefully optimized meta descriptions. They don't click your featured snippet. They ask ChatGPT and get an answer that's been synthesized from training data, and if your content isn't represented in that data, you never existed.

Traditional search: query → results page → click → site → action

AI search: query → synthesized answer → (maybe) click → site → action

The difference is that in traditional search, you're competing to be one of ten blue links. With enough SEO work, you can rank. In AI search, you're competing to be one of maybe three sources the model draws from. Sometimes you don't get any link at all because the model just answers the question from memory.

And people are getting comfortable with this. They don't want 10 links. They want one good answer.

I've spent my career doing SEO and the mental model for AI visibility is different in a few important ways:

Links still matter, but mentions matter more. A high-quality backlink signals authority to Google. But for AI models, being mentioned in multiple trusted sources matters more than the link itself. Think of it as reputation rather than PageRank.

Structured content wins. AI models extract information better from content that's clearly organized. Tables. Bullet points that aren't just keyword-stuffed. Honest comparisons. These formats survive training data better than flowing prose.

Category-defining pages are the new cornerstone content. A page that explains your entire category, not just your product, gets referenced by AI models because it's useful when answering broad questions. "What is X?" pages, comparison pages, and data-driven analyses all punch above their weight in AI visibility.

Traditional on-page SEO still matters but not for the reasons you think. Title tags and headers help AI models understand what your page is about during training. Good metadata helps make your page easier to parse and reference. Not for ranking, for comprehension.

If you have a content team, split your effort. 70% traditional SEO (still a massive channel and will be for years). 30% AI visibility work: structuring your best content for AI comprehension, getting mentions in authoritative sources in your category, and tracking whether your brand actually shows up when people ask AI models about your space.

If you're running a company and you haven't checked your AI visibility yet, do it today. Search for your product category on ChatGPT and Gemini. See if you're mentioned. See who is. You'll either be relieved or you'll have a new problem to solve, and both outcomes are useful.