Search has been quietly self-updating for thirty years. What feels like a sudden AI upheaval is really the same pattern accelerated: publishers across the industry have watched organic traffic drop hard over the past year, and the ones handling it best identified the shift years before it became a headline. Shelley Walsh, a longtime search marketing leader, laid out a practical framework for how brands can survive AI search at a recent industry conference, and it’s worth unpacking because it cuts through both the panic and the hype.

Stop treating rank position as the goal

If your team is still checking rankings daily as the primary success metric, that habit needs retiring. Ranking was the 2016 metric; visibility is the 2026 one.

The core job of search marketing has never actually changed: know your customer, know where they operate, and use content to reach and move them. What has changed is where that interaction happens. Audiences now move fluidly through a multimodal search journey, an AI layer woven through nearly every step, before they ever land on a conclusion.

A number-one ranking no longer guarantees a click, or even guarantees visibility. Advanced Web Ranking found that when an AI Overview expands, the first organic result gets pushed roughly 1,674 pixels down the page, effectively below the fold on most screens. And AI Overviews are just one layer among several; stack ads, carousels, map packs, and image results on top, and a “number one” position can be functionally invisible. Some teams have watched client SERPs shift this dramatically and made a deliberate choice to stop chasing rank as a vanity metric and put that energy into intent- and action-based strategy instead, meeting users where they actually are with a real reason to engage.

Three strategies built to hold up under AI search

None of this means SEO is dead. Technical excellence is still the foundation for being discovered by LLMs, and content is still the foundation of visibility altogether, without it, there’s nothing for AI systems or search engines to find. Three approaches in particular offer real stability through this transition.

1. Build content that resists being flattened into a summary

The core tension in LLM visibility is that you need enough consensus with the rest of the web to be considered trustworthy, but enough genuine differentiation to actually earn inclusion rather than being synthesized away into someone else’s answer. Brands already running original experiments and collecting their own data have a real head start here. SEO consultant Grant Simmons has described this as “golden knowledge”: your data, your experience, your opinion, the parts of your content an LLM cannot generate purely from training data.

In practice, that looks like:

• First-hand formats. Video interviews, live conversations, and lived-experience commentary carry a human perspective that spreads across social, SERPs, and LLM answers precisely because it can’t be fabricated from existing training data.
• Original research and proprietary data. State-of-industry reports and survey-based studies give models something concrete to cite that doesn’t exist anywhere else.
• Opinionated, expert-driven commentary. Analysis from people with real, named experience in a niche is harder to synthesize away than generic explainer content.

Anyone can prompt an LLM to summarize “what is SEO.” Far fewer can offer a community of credible voices, proprietary data, and expert commentary that becomes the actual source an LLM draws from. That kind of content strategy also reduces reliance on any single traffic channel, which matters more every quarter, a point that connects directly to what it actually takes to get cited, and stay cited, in AI search.

2. Design for value-based clicks, not volume

LLM referral traffic is real, if still small in absolute terms. Chartbeat data reported by Press Gazette put ChatGPT’s referral share at roughly 0.02% of publisher traffic, while Conductor’s 2026 benchmarks report puts LLM referral traffic at about 1.08% of website traffic across ten industries. Small percentages of an enormous search volume still represent a meaningful market, and the strategic question is what earns the click rather than the summary.

Ask directly: why would someone click through from an LLM answer instead of stopping at the summary? What does your page offer that the AI response can’t contain? Featured snippets never eliminated clicks entirely, despite years of concern that they would, and the same logic applies here. What tends to drive a click out of an AI answer includes:

• Depth the summary can’t hold. Case studies, implementation detail, and nuance that a three-sentence answer simply can’t fit.
• Credibility and trust signals. Amsive found branded queries paired with an AI Overview actually saw an 18% increase in click-through rate.
• Actionable assets. Tools, calculators, and resources where the underlying intent genuinely can’t be satisfied by a summary.

Separating instant-answer traffic from people who specifically don’t want the quick answer is the distinction that makes a site valuable rather than just visible.

3. Target the SERP real estate AI hasn’t taken over

Google is not disappearing, and it still holds an edge LLMs don’t have: decades of understanding user intent, established audience trust, and a scale of data no competitor can match overnight. Brightedge data shows just over half of queries now trigger an AI Overview, and Conductor found roughly a quarter of the searches it analyzed (21.9 million unique Google searches) triggered one. That means anywhere from half to three-quarters of SERPs still have no AI Overview at all, plenty of room where intent gets satisfied by an actual page visit rather than a generated answer.

Categories that tend to resist AI summarization particularly well include breaking news (moving faster than models can absorb it), branded search (where trust and community pull people to search for you directly), and download-gated resources, still one of the most reliable conversion mechanisms available. AI Overviews may be taking a bite out of raw traffic volume, but they are not taking the traffic that actually converts.

Consensus is the multiplier behind all three strategies

If there’s one lever with outsized impact across everything above, it’s consensus. LLMs generate answers from statistical patterns across their training and grounding data, so a brand or message that shows up consistently across many independent sources is far more likely to surface in an AI answer. Ahrefs found branded web mentions had the strongest correlation with appearing in AI conversations of any factor it tested, stronger than backlinks or on-page signals. Separately, University of Toronto research found LLMs favor earned media from third-party, trusted sources over content published directly on a brand’s own site. You will find more ideas in content writing tips from experts to survive 2026.

Practically, that means layering your best material across Reddit, LinkedIn, YouTube, and relevant industry publications, not hoarding it exclusively on your own domain, builds the cross-source consistency models actually reward. Treat your website as the hub that connects to everywhere else you show up, rather than the only place your content exists. Building that kind of cross-channel visibility into AI recommendations is one of the more durable investments available right now.

What actually changes on a team, not just in theory

The mistake many publishers made early in this shift was treating AI as something happening to them rather than something they could navigate deliberately. The teams that adjusted made a few consistent moves: shifting editorial output toward experience-first formats like interviews, analysis, and original research; moving revenue models away from pure programmatic dependence and toward asset-based partnerships; and prioritizing owned, direct audience growth as the top metric instead of raw session counts, which also changes what gets reported to clients and leadership when traffic dashboards no longer tell the full story.

If a team is still running the same SEO playbook it used in 2020, the fix isn’t panic, it’s revisiting who the audience actually is, where they spend attention now, and who’s competing for that attention across every discovery surface, not just Google’s results page. Brands that survive AI search in 2026 will be the ones treating visibility, not rank, as the metric that counts.

Frequently asked questions

Is SEO still relevant with AI search growing?

Yes. Technical SEO fundamentals remain the basis for being discovered by LLMs, and content is still the foundation of visibility across every surface, including AI-generated answers. What has changed is the metric that matters: visibility across a multimodal journey, not rank position alone.

How much traffic do LLMs actually send to websites right now?

Estimates vary by source. Chartbeat data cited by Press Gazette put ChatGPT referrals at around 0.02% of publisher traffic, while Conductor’s 2026 benchmarks report puts total LLM referral traffic closer to 1.08% across the industries it studied. Both are small percentages of an enormous overall search volume.

What content is hardest for AI to summarize away?

First-hand formats like interviews and lived-experience commentary, original proprietary research, and named expert opinion tend to hold up best, because none of it can be reconstructed purely from an LLM’s existing training data.

What is “consensus” in the context of AI search visibility?

It’s the degree to which a brand or message appears consistently across many independent sources, not just its own website. Research from Ahrefs and the University of Toronto both point to consistent, third-party mentions as a stronger driver of AI citation than owned-site content alone.