The New Search Funnel: How Discovery Actually Works Now
The new search funnel is decentralized: people move between AI assistants, search engines, YouTube, Reddit, and review platforms in whatever order makes sense to them, instead of following one linear path. The old model, Google Search, then your website, then a lead, then a customer, worked for twenty years. It’s not how people find businesses anymore, and clinging to it is why so many teams are confused about their own numbers right now.
The new search funnel is decentralized: people move between AI assistants, search engines, YouTube, Reddit, and review platforms in whatever order makes sense to them, instead of following one linear path from Google to a website to a lead. By the time someone lands on your actual website, they may have already formed an opinion about you based on what an AI assistant told them somewhere else.
- No single platform owns the discovery journey anymore, Google and ChatGPT lead in scale, but Claude, Perplexity, Gemini, and niche community engines all carry real share of the conversation.
- Stable rankings with falling traffic isn’t a contradiction, it’s the new funnel working as designed, AI Overviews and chat assistants increasingly answer directly without requiring a click.
- The practical shift is moving from optimizing for rankings and clicks to optimizing for presence everywhere the question gets asked, treating entity signals and structured data as seriously as Google rankings.
The Five Doors of Discovery
Discovery now runs through five surfaces, in whatever order the buyer chooses:
- AI assistants. ChatGPT, Claude, Gemini, and Perplexity answer questions directly, often without a search engine ever being involved.
- Search engines with AI answers. In Pew’s tracked panel, 18% of Google searches opened with an AI summary instead of a list of links.
- YouTube. Video is where a large share of how to and evaluation questions now resolve.
- Reddit and community platforms. Buyers check what real people say before they trust what brands say.
- Review platforms. The trust check that happens before, or instead of, a website visit.
How Did the Old Search Funnel Work?
The old model was clean because it had one entry point. Someone searches Google, clicks a blue link, lands on your site, and either converts or doesn’t. Every tool you’ve ever used to measure marketing performance, most CRMs, most attribution models, was built around that single door.
That door is still open. It’s just no longer the only one, and for a lot of queries, it’s not even the first one people walk through anymore.
What Does the New Search Funnel Look Like?
Today someone might ask an AI assistant a question directly and never touch a search engine at all. They might watch a YouTube video, read a Reddit thread, check review platforms, or get an answer summarized inside a search results page with a citation instead of a click. They bounce between these surfaces in whatever order makes sense to them, and by the time they land on your actual website, if they land there at all, they may have already formed an opinion about you based on what an AI assistant told them.
No single platform owns this journey. Google and ChatGPT lead in scale, but Claude, Perplexity, Gemini, and a growing list of niche and community engines all carry real share of the conversation, and most dashboards still can’t see any of it.
The scale of the shift is measurable. In Pew Research Center’s tracked browsing panel of 900 US adults, 18% of Google searches triggered an AI summary. That is a general finding from one measured panel, and it means roughly one search in five now opens with an answer instead of a list of doors.
Old SEO vs New SEO: What Changed?
- Old SEO meant earning the top spot on one results page. New SEO means earning presence across every surface a question travels through before it resolves.
- Old SEO meant getting the click. New SEO means getting mentioned before the click ever happens, sometimes instead of it happening at all.
- Old SEO meant winning Google. New SEO means being the answer wherever people happen to ask, the shift Neil Patel calls search everywhere optimization.
Why your rankings can hold steady while your traffic falls
This is the pattern we hear from almost every new client: “our rankings are stable, but traffic is down.” That’s not a contradiction, it’s the new funnel working exactly as designed. AI Overviews and chat assistants increasingly answer the question directly on the results page or inside the conversation itself. The ranking is still earning its position. It’s just not earning the same volume of raw clicks it used to, because fewer people need to leave the page to get their answer. Ahrefs’ click study measured the effect directly: when an AI Overview appears, clickthrough to the top organic result drops sharply.
The visibility didn’t disappear. It moved from a click you can count to a citation or mention you have to go looking for. That’s a real shift in what “performance” even means, and it’s one most reporting tools haven’t caught up to yet.
Our own Search Console data shows the same funnel shape from the inside. Across one 28 day window, 11 of the 13 clicks the site earned went to brand and trust pages, the homepage, the about page, the founder page, while informational pages collected thousands of impressions and almost no clicks. People were not arriving to discover us. They arrived already having decided to check us out, which means the discovery step happened somewhere we could not see. That is an internal observation from a single site’s data, not a study, but it is the new funnel leaving footprints.
What this means for how you should be thinking about discovery
The practical shift is moving from optimizing for rankings and clicks to optimizing for presence everywhere the question gets asked. That means treating your entity signals, structured data, and cross platform consistency as seriously as you’ve always treated your Google rankings, because AI assistants are reading and citing from that same foundation, just through a different door.
We go deeper on how AI systems actually decide what to cite in how AI search engines build answers, on why stable rankings can still come with falling traffic in zero-click search explained, and on what to measure now that clicks aren’t the whole picture in how to measure GEO.
If you want a clear read on where your brand actually shows up across this fragmented landscape, that’s exactly what we map out when you book a call.
FAQ
Is the traditional search funnel dead?
Not dead, but no longer the only path. Google search still matters, but discovery now also happens through AI assistants, YouTube, Reddit, and review platforms, often before someone ever reaches your website.
Why do multiple platforms matter instead of just Google?
No single platform owns discovery anymore. Google and ChatGPT lead in scale, but Claude, Perplexity, Gemini, and niche community engines all carry real share of how people find and evaluate brands.
How do I optimize for a funnel I can’t fully see?
Focus on the underlying signals that carry across every surface: entity clarity, structured data, and cross platform consistency. These are what AI assistants and search engines both read from, regardless of which door someone walks through.
Does the new funnel mean classic SEO is a waste of money?
No. AI Overviews sit on top of Google’s index, and most AI assistants retrieve from pages search engines already rank well. Classic SEO is the foundation the new surfaces read from. What changed is that ranking alone no longer captures the value, since the answer often resolves before the click.
How can I measure discovery that never shows up in my analytics?
Track the downstream traces. Branded search volume rises when people hear about you elsewhere and look you up. Testing a fixed set of buyer questions across ChatGPT, Claude, Perplexity, and Google each month shows whether you are being named. Direct traffic and mention volume fill in the rest of the picture.
Where should a small business start in the new funnel?
Start where your buyers already ask questions. Run your three most important customer questions through the major AI assistants and note who gets recommended. Fix your entity basics next, consistent naming, structured data, a clear About page, then build presence on the one platform outside Google where your customers actually spend time.
What To Do Next
Three moves, in order:
- Run your three most important customer questions through ChatGPT, Claude, Perplexity, and Google, and write down who gets named. That is your real funnel position, and it takes twenty minutes.
- Fix cross platform consistency so your name, description, and core facts read identically everywhere a model might look.
- Pick the one surface outside Google where your buyers already spend time and build genuine presence there, instead of spreading thin across all of them at once.
How do we know this works? The Pew panel above shows the click path thinning at the top of the funnel, our own Search Console data shows visitors arriving already decided, and we publish exactly how we measure AI visibility on our results page.
Want us to do the work for you? We build content hubs for brands and implement everything we teach here, across every surface in this article.
Every statistic above links to its original source where it first appears. We label numbers by type: a general finding comes from published third party research, a sample result comes from a defined study sample, a client result comes from Winsignia client work, and an internal observation comes from our own audits and site data. This article draws on Pew Research Center’s tracked browsing panel of 900 US adults (general finding), Ahrefs’ AI Overview click study (general finding), and an internal observation from our own Search Console data. Where a practice is documented by Google directly, we say so. Where it reflects industry practice ahead of official guidance, we treat it as informed inference rather than settled fact.
Scott Fishman is Director of Growth at Winsignia, where he builds AI search visibility systems for service businesses and expert brands. He built and maintains the content hubs behind stpetemarathon.com and unitedsportsassociation.org, and winsignia.io runs on the same system he sells. He is the author of the book EFFORT. This article was reviewed and updated on August 25, 2026.
Related Reading
Why Your Rankings Are Stable But Your Traffic Isn’t
From Clicks to Citations: The Metrics That Actually Matter
What Is Generative Engine Optimization (GEO)?
SEO, AEO, and GEO Aren’t Three Things Anymore
AI Search Statistics 2026
AI Search Statistics 2026: SEO, AEO, and GEO Data
Want to Be the Brand AI Recommends?
Book a call and we’ll map what it takes to get there for your business.