A few months ago I was helping a friend audit his content site β€” a solid niche blog he'd been building for three years, really good stuff, genuinely helpful. His rankings were fine. His impressions in Search Console were climbing. But his traffic? Down 34% year over year. Not tanking, but a slow, steady bleed that he couldn't pin on anything specific.

Then we looked at his AI Mode query data. Turns out, a huge chunk of his highest-impression keywords had started triggering AI Mode responses. Google was answering the questions directly, completely, and confidently β€” and users just… didn't need to click anywhere. His rankings hadn't changed. His CTR had been quietly gutted.

This is the story of 2026 SEO for a lot of sites. And the number behind it is 93% β€” that's the share of Google AI Mode searches that end without a single click to any website, according to Semrush data. For comparison, AI Overviews (the older, less immersive format) have a zero-click rate of about 43%. AI Mode makes AI Overviews look like a traffic firehose.

So what do you actually do about this? Panic is one option. I've seen a lot of that on SEO Twitter lately. But there's also a smarter path β€” and it involves understanding why some sites are still winning clicks even in AI Mode results, and what separates them from the ones getting quietly wiped out.

What Google AI Mode Actually Is (And Why It's Different)

Most people conflate AI Mode with AI Overviews, but they're not the same thing and the distinction matters. AI Overviews are that summary box that appears above the regular blue links β€” they answer the query and then show you sources below. You can still scroll past them and click a regular result.

AI Mode is something else. It's a dedicated conversational search experience β€” essentially a chat interface built into Google Search. When a user switches to AI Mode, they get a full, multi-paragraph response to their query that's generated in real time. There are usually one or two inline citations, but the experience is designed to keep you in the conversation, not send you away to a website.

⚠️
The Core Problem AI Mode doesn't just reduce clicks β€” it replaces the need to click. Users get a complete, conversational answer right there. The few sites that do get cited receive citation traffic, but it's a fraction of what traditional ranking positions deliver.

Here's what makes this particularly rough for content publishers: AI Mode is getting more popular fast. Google has been nudging users toward it across mobile especially, and the "AI Mode" tab is increasingly prominent in the search interface. Some queries that would have triggered a normal SERP a year ago are now defaulting to AI Mode for a growing percentage of users.

93%
AI Mode searches with zero clicks to any website
43%
AI Overviews zero-click rate (for comparison)
58.5%
All Google searches ending without a click in 2026

Which Queries Are Getting Eaten Alive

Not all keywords are equally at risk. After spending a lot of time with Search Console data and watching AI Mode behavior across different niches, I've noticed some clear patterns.

Queries that AI Mode is absolutely devouring:

  • Factual definitions and explanations β€” "what is X", "how does Y work", "explain Z". Google answers these perfectly without needing to cite anyone.
  • Simple how-to queries β€” "how to add a canonical tag", "how to compress an image". If the answer fits in 4 sentences, AI Mode gives 4 sentences.
  • Comparison queries β€” "X vs Y" used to be great content fodder. Now AI Mode just builds a comparison table on the fly.
  • Price and availability questions β€” "how much does X cost", "is Y free". These are gone entirely to AI answers.
  • News summaries β€” What happened with the Google algorithm this week? AI Mode will tell you without sending you anywhere.

Queries that still drive real clicks:

  • Tool and software searches β€” People searching for actual tools need to visit the tool. You can't use a redirect chain checker inside a Google AI answer.
  • Complex, opinion-heavy topics β€” AI Mode hedges on controversial or nuanced stuff. Humans still want a human take.
  • Case studies and original data β€” If you have something Google can't synthesize from existing content, you get cited and clicked.
  • Deep technical how-tos with specifics β€” "How to fix a redirect chain in a specific CMS" beats "what is a redirect chain" by a mile.
  • Local and transactional intent β€” "best Italian restaurant near me", "buy X near Y" β€” these still convert to clicks.
πŸ’‘
The Pattern AI Mode kills clicks on queries where the answer is the destination. It spares queries where the website itself is the destination β€” because you need to actually use something, do something, or read an original perspective that can't be summarized away.

The 7% That Still Get Clicked β€” How Citations Actually Work in AI Mode

That 7% of clicks that do happen in AI Mode aren't random. There's a pattern to which sites get cited and which get silently used as training data that the user never sees.

From what I've observed β€” and this matches what others in the SEO community have been reporting β€” AI Mode tends to cite sources that have a few things in common:

They're cited elsewhere on the web, not just ranking well. AI Mode seems to pull from sources that have genuine authority signals β€” sites that are referenced by journalists, linked to from authoritative domains, mentioned in other AI answers, and have brand presence beyond just their Google rankings. A site that ranks #1 via pure SEO optimization but is never cited or mentioned anywhere else tends to get its content absorbed without attribution.

They offer something original. Original research, unique datasets, a distinctive point of view that Google can't synthesize from averaging five similar articles. If your content could have been written by any of ten competing sites with minor variations, AI Mode will just generate the average version and nobody gets cited.

Their entity clarity is solid. This sounds abstract but it's practical: does Google understand who you are, what you cover, and why you're authoritative on this topic? Sites with good Knowledge Graph presence β€” proper structured data, consistent brand mentions, clear author entities β€” get cited more than amorphous content farms with no clear identity.

"The sites winning in AI Mode aren't ranking their way to citations β€” they're building their way there. It's a brand and reputation game dressed up as an SEO game."

What To Actually Do: 5 Concrete Moves

Enough diagnosis. Here's what I'd actually change in my content strategy right now if I were running a site that's starting to feel this zero-click squeeze.

1

Audit Which Keywords Are Already in AI Mode

Before anything else, figure out your exposure. Pull your top 100 keywords by traffic from Search Console and manually check which ones now trigger AI Mode. This tells you where your traffic is at risk. There's no automated tool that does this perfectly yet, but spot-checking your biggest volume terms takes maybe 45 minutes and gives you a real picture of the threat.

2

Rewrite Your Informational Content to Be Opinionated

Generic "what is X" and "how does Y work" content is now largely pointless β€” AI Mode handles it better than you can. But opinionated takes are different. "What is a redirect chain" gets eaten. "Why I think most SEOs are handling redirect chains wrong (and the right approach for 2026)" might survive because it's a perspective, not just information. Retrofit your high-traffic informational pieces with genuine opinions, original examples from your own experience, and takes that AI can't average into existence.

3

Build More Content Around Your Tools or Products

If your site has something people need to use β€” a tool, a calculator, a generator, a template download β€” that's your moat. Content that lives alongside usable things drives clicks because reading about it isn't enough; they have to come to you. Focus content efforts around the things only you can provide, not information anyone can provide.

4

Do Something Original That Can Be Cited

Run a survey. Publish a dataset. Do an experiment that nobody else has done and write it up with real numbers. This type of content gets cited in AI Mode because it's genuinely irreplaceable β€” Google can't synthesize original data, it can only reference it. One good original study can earn citations across dozens of AI Mode responses over months.

5

Strengthen Your Entity Signals

Make sure Google knows who you are. That means: consistent brand name and author names with structured data markup, active presence on platforms where entities get reinforced (LinkedIn, niche directories, press mentions), and a clear topical focus so you're recognized as an authority on specific subjects rather than a generalist. This is unsexy work but it directly influences citation likelihood in AI Mode responses.

πŸ€– See Where You Stand in AI Search Right Now

RankSorcery's AI Search Visibility tool shows you how your site is performing across AI-powered search results β€” and what's blocking you from getting cited. No login required.

Check My AI Visibility β†’

What Not to Do (Common Panic Moves That Backfire)

A few responses to AI Mode's rise that I think are mistakes, even though I see them recommended a lot:

Don't just add more words. There's a theory floating around that longer, more comprehensive content gets cited more in AI Mode. Maybe β€” but the causality is backwards. Comprehensive content earns citations because it's comprehensive and authoritative, not because it's long. Adding 2,000 words of padding to a 1,000-word post won't change your AI Mode visibility at all.

Don't abandon informational content entirely. Even if clicks from informational queries are down, that content still serves a brand awareness function. Someone who reads an AI Mode answer that cites you might look up your brand directly later. Assisted attribution from AI citations is real, even if it's hard to measure.

Don't chase AI Mode citations with thin Q&A content. Some people are writing FAQ-style pages specifically to get picked up as AI Mode sources. This might work short-term but it's the same trap as optimizing for featured snippets circa 2019 β€” Google gets better at answering those questions itself and you get squeezed out again.

🎯
The Long-Term Play The sites that will thrive in an AI Mode world are the ones that have a reason to exist beyond ranking. A genuine tool, an original voice, unique data, or a community. If your site's only value proposition is "I rank for these keywords," that's the value proposition that AI Mode eliminates.

Actually Measuring Your AI Search Visibility

One of the frustrating things about AI Mode is that Google Search Console doesn't break it out cleanly from regular search. Your impressions and clicks look like they're coming from the same place, but the intent and behavior behind them are completely different.

What I'd track specifically:

  • CTR trends on your top informational keywords β€” a falling CTR with stable impressions is the AI Mode signature
  • Branded search volume β€” if AI Mode citations are driving brand awareness, you should see a slow climb in brand searches over time
  • Direct and referral traffic as a percentage of total β€” not all value flows through clicks from search, and this signals whether you're building real brand gravity
  • Which competitor sites are getting cited in AI Mode answers for your target topics β€” understanding who Google trusts helps you understand what you need to become

That last one is something RankSorcery's AI Search Visibility tool can help with β€” it analyzes how AI search engines are picking up and citing content in your space, so you can see where the citations are going and what type of content is earning them. Worth running before you overhaul your content strategy based on guesswork.

The Bigger Picture: This Is a Shift, Not a Death

I want to be clear that I'm not in the "SEO is dead" camp. It's not. But it's also undeniably different, and the 93% zero-click stat in AI Mode is a number you can't just optimize your way around with better title tags.

What we're really watching is a shift in what "visibility" means. For most of search history, visibility meant ranking position. Now visibility means being cited, being known, being the entity that AI surfaces when someone asks about your topic area. That's a slower, harder thing to build β€” it's brand-building dressed up as SEO.

The practical upshot is that the gap between sites that have real brand authority and sites that are purely SEO-optimized is going to keep widening. If you've been building something real β€” an audience, a reputation, original tools or research β€” AI Mode is annoying but survivable. If you've been building a ranking machine with no underlying authority, the squeeze is going to get worse every quarter.

The friend with the niche blog I mentioned at the start? We ended up rebuilding his content strategy around his actual expertise β€” he's a former product manager, and he had genuinely useful experience in his niche that none of his competitors could replicate. Six weeks later, his CTR on the updated posts is meaningfully better, and one of his original case studies got cited in an AI Mode response. It's not a magic fix. But it's a real direction.

Adapt toward things only you can offer. That's always been the right SEO advice. AI Mode just made it urgent.

JR

James Reyes β€” RankSorcery

James has been doing SEO for longer than he'd like to admit. He runs RankSorcery and writes about the parts of search that don't make it into the standard playbooks. He's been wrong about a few predictions. He's been embarrassingly right about others.