Google Isn't Sending You Traffic Anymore. Here's How to Get Found by AI Instead

If you've checked your Google Analytics lately and felt a small knot in your stomach, you're not imagining things. Traffic is down. Rankings look the same. Nothing in Search Console screams "penalty." And yet, month over month, fewer people are landing on your site.

You're not doing anything wrong. Search itself has changed.

What's actually happening

For almost two decades, the deal was simple. You wrote something useful, Google ranked it, and people clicked through to read it. That deal is quietly falling apart.

Google now answers most questions directly on the results page through AI Overviews, and it's not the only one doing this. ChatGPT, Perplexity, and Gemini are increasingly where people go to ask a question and get an answer without ever opening a browser tab. Search hasn't gone away. Clicking has.

Some of the numbers floating around are genuinely rough. Publishers as a group have reported organic traffic drops in the double digits over the past year, and a handful of well known companies have talked openly about losing well over half their search traffic since AI Overviews rolled out widely. Even when a page holds its ranking position, the click through rate on that position has fallen sharply once an AI summary shows up above it.

So no, you didn't lose your SEO edge. The page where your edge used to matter just got smaller.

Why this isn't the "SEO is dead" story you've heard before

Here's the part most doom and gloom posts skip over. This shift isn't only taking traffic away. It's also handing out a new kind of advantage to a smaller group of brands, and that advantage is bigger than what was lost.

Brands that get cited inside an AI Overview or referenced in a ChatGPT answer tend to see noticeably higher click through rates than brands sitting just below them in the rankings, uncited. And the people who do click after seeing your brand mentioned by an AI tool arrive already half convinced. They're not casually browsing. They asked a specific question, got a shortlist of answers, and picked you. That's a warmer lead than most blog traffic has ever produced.

In other words, the traffic pool got smaller, but the water got a lot more valuable. The businesses that figure out how to get cited are going to pull ahead of the ones still writing for 2019 style search rankings.

The real fix: stop optimizing for rankings, start optimizing for citations

This is where a newer idea comes in, often called Generative Engine Optimization, or GEO for short. Some people call it Answer Engine Optimization instead. The name matters less than the shift in thinking behind it.

Traditional SEO asks: how do I rank higher than the next result?

GEO asks: how do I become the source an AI model trusts enough to quote?

Those are different questions with different answers. Here's what actually moves the needle.

Write things nobody else can copy. An AI model can summarize "how to register an LLC" because five hundred other sites have already written the exact same generic explainer. It cannot summarize the fact that you've personally walked two hundred clients through that exact process and know where they usually get stuck. Original experience, real numbers from your own work, and specific client stories are much harder for a machine to flatten into a generic answer.

Answer the question in the first two sentences. AI models tend to pull from content that states the answer plainly near the top of a section, then backs it up below. Bury your best insight under three paragraphs of throat clearing and you're making it easy to skip.

Let the bots in. Check your robots.txt file. If you're accidentally blocking GPTBot, ClaudeBot, or similar crawlers, none of this matters because the AI can't read your page in the first place.

Build a real reputation outside your own website. AI models weigh how often you're mentioned across the internet, not just what you say about yourself. Getting quoted in industry publications, showing up in review sites, earning genuine backlinks, and having a named author with real credentials all feed into whether a model trusts you enough to cite you.

Go deep on a topic instead of scattering thin posts everywhere. A handful of well connected pages that thoroughly cover every angle of a subject tend to outperform a pile of shallow, disconnected posts. Think of it as building a small library on your topic rather than a stack of pamphlets.

Add something a summary can't replace. A calculator, a quiz, an interactive comparison tool, original survey data. These give people a reason to actually visit your site instead of just reading a synthesized answer about it.

How to know if this is actually happening to you

Before you overhaul your whole content strategy, check whether AI Overviews are really the cause of your drop, or whether something else is going on.

Open Search Console and compare impressions against clicks over the past few months. If impressions are holding steady or even climbing while clicks are sliding, that's a strong sign your rankings are fine but something on the results page is intercepting the click.

Then look at which pages are losing the most ground. Simple explainer content, how to guides, and definition style posts tend to get hit hardest, since those are exactly the queries AI Overviews are built to answer directly. Pricing pages, comparison pages, and anything with a strong buying intent usually hold up better.

It's also worth setting up a way to track AI referral traffic separately in your analytics. It won't be a huge number yet for most sites, but watching whether it grows month over month tells you if your citation efforts are working.

What to actually do this month

You don't need a total content overhaul to start. Pick your five highest traffic pages that have dropped the most and work through them one at a time.

Add a section with an original stat, case result, or personal observation that nobody else's version of this article has. Tighten the opening of each section so the answer shows up immediately instead of after a long wind up. Check that your author has a real name and bio attached. Look for one or two places where you could add a comparison table, a short calculator, or some kind of interactive element.

Then leave it alone for a month and check Search Console again. This isn't a switch you flip once. It's closer to compound interest. The pages that get cited early tend to keep getting cited, because AI models return to sources they've already found useful.

The traffic model built on cheap clicks is winding down. What's replacing it rewards fewer, better sources. That's genuinely good news if you're willing to be one of them.