LLMs Are Changing SEO. Here Is What Matters

Large language models have changed how a meaningful portion of search queries get answered. Not every query, not every user, and not every category equally, but enough that treating SEO strategy as a static discipline is now a commercial risk. The core question is not whether LLMs matter to SEO. It is which parts of your SEO strategy need to change, which parts hold, and which parts of the conventional wisdom were already wrong before AI entered the picture.

The honest answer is that LLMs have accelerated some shifts that were already underway, created genuinely new challenges around visibility and attribution, and exposed how fragile some SEO programmes were to begin with. If your strategy was built on volume, thin content, and rank tracking as a proxy for business performance, it was already on borrowed time.

What LLMs Are Actually Doing to Search Behaviour

The mechanism worth understanding is not the technology itself but the behavioural shift it creates. When a user gets a synthesised answer directly in a search interface, whether through Google’s AI Overviews, Bing’s Copilot integration, or a standalone tool like ChatGPT or Perplexity, the probability of a click to your site drops. Not to zero, but meaningfully. For informational queries, the ones that have historically driven the top of most content funnels, this compression is already visible in traffic data for many sites.

This is not a catastrophe. It is a recalibration. Informational traffic that never converted was always of limited commercial value. The more important question is what happens at the middle and bottom of the funnel, where purchase intent is higher and where LLMs are less likely to fully resolve the query without a click. That is where SEO still has clear commercial leverage, and where strategy should be concentrated.

Semrush has published useful analysis on how Google’s AI Mode is affecting SEO and organic visibility, and it is worth reading if you want data to frame internal conversations. The short version: impact varies significantly by query type, and the sites holding up best are those with strong topical authority and clear entity signals.

The Attribution Problem Gets Worse Before It Gets Better

I have spent a lot of time in rooms where people are arguing about marketing attribution. Running agencies across 30 industries, managing substantial ad spend, I have seen how quickly measurement conversations become political. Everyone wants the credit, no one wants the cost. LLMs make this problem structurally harder.

When a user asks an LLM a question, gets a partial answer that mentions your brand, then searches for you directly three days later, that conversion shows up as direct or brand search in your analytics. The LLM interaction is invisible. This is not a new problem, it is an amplified version of the dark funnel challenge that has existed since content marketing became mainstream. But the scale at which LLMs operate means the gap between what your analytics show and what is actually influencing purchase decisions is widening.

My view on measurement has always been that analytics tools give you a perspective on reality, not reality itself. The businesses that get this right are the ones that invest in honest approximation: brand tracking, incrementality testing, and qualitative research alongside the quantitative. If you are still treating last-click attribution as the truth, LLM-era search will make that position increasingly indefensible.

If you want a broader grounding in how SEO strategy should be structured before layering in LLM considerations, the Complete SEO Strategy hub covers the full picture, from technical foundations to content and measurement.

Why Topical Authority Matters More Than It Did

LLMs do not rank pages. They synthesise information from sources they have been trained on or, in retrieval-augmented systems, from sources they can access in real time. The sources that get cited, recommended, or drawn upon tend to share common characteristics: they are specific, they demonstrate genuine expertise, they are consistent across multiple pieces of content on a topic, and they are associated with identifiable authors or organisations with credible track records.

This is essentially a description of topical authority, and it maps closely to what Google has been rewarding through its quality guidelines for years. The difference now is that the stakes are higher. If an LLM synthesises an answer to a query in your category and draws on three or four sources, being one of those sources is worth more than ranking fifth for a keyword that gets a featured snippet. Presence in the synthesis is the new first page.

Building topical authority is not complicated in principle, though it is genuinely hard in practice. It requires consistent, specific, expert-level content across a defined subject area. It requires real authorship, meaning content associated with people who have demonstrable credentials in the field. And it requires patience, because authority accrues over time and does not respond to shortcuts.

I judged the Effie Awards for several years, reviewing campaigns that were trying to demonstrate measurable business effectiveness. One pattern that appeared consistently in the strongest entries was a clear point of view, a distinctive position that the brand owned over time. The same principle applies to content. The sites that LLMs draw on are the ones that have a clear, consistent, authoritative voice on a topic, not the ones that covered every keyword variation with interchangeable 800-word articles.

Structured Data and Entity Clarity Are No Longer Optional

One of the more practical implications of LLM-influenced search is that structured data and entity clarity have moved from best practice to baseline requirement. LLMs and search engines alike benefit from clear signals about who you are, what you do, and how your content relates to specific topics and entities.

This means schema markup implemented correctly and completely, not just Article and FAQ schema on every page regardless of relevance. It means your organisation’s entity is clearly defined: consistent NAP data if you are a local business, a well-maintained Knowledge Panel if you are a brand, clear author profiles with credentials if you are a publisher. It means internal linking that reflects genuine topical relationships rather than just passing PageRank.

The Moz team has done solid work explaining how brand SEO strategy intersects with entity signals, and it is a useful frame for thinking about how LLMs decide which sources to trust. Brand is not just a marketing concept in this context. It is a trust signal that search systems, including LLMs, use to evaluate source reliability.

For teams managing more complex technical environments, including headless architectures where content delivery and rendering are decoupled, the structural implications are worth examining carefully. Moz has also covered the SEO considerations specific to headless setups, which become more consequential when you are trying to ensure LLMs can reliably access and interpret your content.

Content Volume Strategies Are Facing a Reckoning

I have seen content strategies built almost entirely on volume. The logic was straightforward: more pages, more keywords, more traffic. For a period, it worked well enough that it became the default playbook. Agencies sold it, in-house teams adopted it, and the economics made sense when the cost of content production was low and the traffic returns were reliable.

LLMs have broken this model at both ends. On the supply side, the cost of generating content at scale is now effectively zero, which means volume alone has no competitive value. On the demand side, LLMs are absorbing the informational queries that thin content was designed to capture, returning synthesised answers without requiring a click. The sites that built their organic presence on high-volume, low-depth content are seeing the most significant traffic erosion, and there is no quick fix.

The response is not to produce less content. It is to produce content that has a reason to exist beyond keyword coverage. Content that reflects genuine expertise, takes a specific position, provides analysis that cannot be replicated by a language model drawing on generic sources, or addresses a question with enough specificity and depth that it earns citation rather than replacement. Semrush’s overview of what a well-structured SEO strategy looks like is a reasonable reference point for teams rebuilding their approach from the content layer up.

Brand Search Is Now a Strategic Asset, Not a Vanity Metric

Early in my career at lastminute.com, I ran a paid search campaign for a music festival and watched six figures of revenue come in within roughly a day. The campaign itself was not complicated. What made it work was that the brand had enough recognition that when people searched with intent, the conversion was almost frictionless. The demand existed. The brand was trusted. The campaign just connected the two.

That experience shaped how I think about brand and performance as connected rather than competing. In the LLM era, this connection becomes more important. When LLMs generate answers that mention or recommend your brand, they are feeding brand search volume. When users who have encountered your brand through an LLM interaction later search for you directly, that branded traffic is attributable to an LLM touchpoint that your analytics will never record.

This means brand investment, the kind that drives awareness and preference rather than immediate conversion, has a measurable downstream effect on SEO performance even if the measurement chain is imperfect. Teams that have historically treated brand and SEO as separate budget lines should be examining whether that separation still makes sense.

What Needs to Change in How SEO Is Reported

If there is one operational change that LLMs demand more urgently than any technical SEO adjustment, it is in how SEO performance is reported and evaluated internally. Rank tracking as a primary KPI was already a weak proxy for business impact. In an environment where a top-three ranking for an informational query might generate near-zero clicks because an AI Overview is absorbing the demand, it becomes close to meaningless.

The metrics that matter are the ones connected to commercial outcomes: qualified organic traffic (not just total traffic), conversion rates from organic, revenue or pipeline influenced by organic, and brand search volume as an indicator of awareness and trust. These are harder to report and harder to defend in a board meeting, but they are the ones that tell you whether your SEO programme is doing anything commercially useful.

I have sat through enough marketing reviews where the numbers looked impressive and the business was flat to be sceptical of any reporting framework that prioritises activity metrics over outcome metrics. LLMs have given SEO teams an opportunity to reset this conversation. Use it.

HubSpot’s work on building SEO strategies that account for a broader range of user intent and behaviour is worth reading as a frame for thinking about how audience-first thinking should inform both content and measurement. The same instinct that drives inclusive SEO, understanding who is actually searching and why, is the instinct that produces better measurement frameworks.

The Opportunity That Most Teams Are Missing

Amid all the discussion about what LLMs are taking away from SEO, there is a genuine opportunity that receives less attention. LLMs are not just consuming content. They are also creating demand. When a user asks an LLM a question and gets a partial answer, they often follow up with a more specific search. When an LLM recommends a category of product or service, it creates purchase intent that then flows into search. The brands and content sources that are well-positioned in LLM outputs are benefiting from a new form of top-of-funnel exposure.

Getting into that position requires the same things that have always driven SEO success at its best: genuine expertise, clear communication, consistent publishing, and a brand that people trust. The tactics change. The fundamentals do not. Teams that understand this will spend less time chasing algorithm updates and more time building the kind of credibility that survives them.

The broader SEO strategy picture, including how to build programmes that hold up across algorithm shifts and channel changes, is covered in depth across the Complete SEO Strategy hub. If you are rebuilding your approach in response to LLM-era search, that is a useful place to pressure-test your thinking.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions