AI Is Reshaping SEO in 2025: What Agencies Need to Know

Artificial‑intelligence‑powered search has transformed the way people discover and research businesses. 2025 is the year when search engines are no longer the only gatekeepers — large‑language models (LLMs) like ChatGPT, Claude and Perplexity have become research assistants that pull answers directly from web content and combine them with training‑data patterns. This shift is forcing SEO agencies to rethink strategy, metrics and how they deliver value to clients.

The Search Revolution: From Keywords to Questions

LLMs and Google’s AI Overviews are changing user behaviour. According to Planetary CEO Josh Gross, what’s happening in search right now is “the most fundamental shift agencies have faced in over a decade”. Traditional SEO tactics focused on ranking pages for individual keywords. Today, people often ask conversational questions (“What’s the best laundry detergent for pet stains?”), and AI tools like ChatGPT and Google’s AI Overviews pull answers from multiple sources and present them as succinct summaries. Gross argues that this means agencies must optimise content for AI consumption rather than for a single keyword.

• AI‑driven search uses natural‑language queries and looks for direct answers. Content must be organized around question‑and‑answer formats and be structured so AI systems can extract fact. It’s not enough to stuff pages with keywords or tack on a FAQ section; agencies need to fundamentally rethink how information is created and presented.
• Josh Gross splits SEO into two parts: technical (site architecture, structured data and alt‑tags) and content‑focused. While technical hygiene remains critical, the content layer “requires a complete rethink” because agencies need to optimise for AI consumption and question‑and‑answer formats.
• This change also affects measurement. Traditional metrics such as click‑through rate and rankings no longer capture how people discover brands when AI tools answer questions without sending users to websites. Gross notes that agencies “lose all that valuable data” when AI tools provide answers directly. New tools like OtterlyAI and Peec attempt to track AI‑based search appearances

SEO Fundamentals Still Matter

A June 2025 guide by SEO expert Olga Zarr makes it clear that marketers shouldn’t panic; AI SEO is still SEO. In fact, “AI SEO optimization isn’t new – it’s still SEO, just with some AI features layered on top”seosly.com. Google still controls more than 90 % of search traffic, while all AI chatbots combined account for <3 %seosly.com. Key takeaways from Zarr’s research include:

Classic SEO principle	Why it still matters	Evidence
Technical SEO	Search engines and AI tools can’t cite pages they can’t crawl. Ensure proper architecture, load speed, mobile‑friendliness and alt‑tags	The guide stresses that technical SEO is a must: “If AI can’t crawl it, it won’t cite”
E‑E‑A‑T and trust signals	AI uses the same trust signals as Google: expertise, experience, authoritativeness and trustworthiness. Quality and credibility are more important than ever
Structured content & schema	Organised, answer‑ready content and structured data help AI understand and cite information
Brand building	AI overviews cite names they recognise. Building a strong brand increases visibility when AI synthesises results

Zarr also notes that citations may be more important than rankings — top positions matter little if AI ignores your page. Unlinked mentions now count because LLMs “see your name even without a backlink”. This means that PR, forums like Reddit and contributions across the web become key signals for AI.

Optimising for AI: From GEO to AAIO

The emerging discipline of AI SEO, also called Generative Engine Optimisation (GEO) or Agentic AI Optimisation (AAIO), focuses on making content machine‑readable. According to the How the F*ck report on Top AI SEO Agencies in 2025, AI SEO is about shifting focus from search engines to large language models. These models “don’t rank content the same way Google does; instead, they read, interpret and generate answers based on patterns and associations”. To show up in their answers, content must be LLM‑readable and contextually strong. The report recommends:

• Make content easy for LLMs to read and cite – Use clear headings, concise explanations and entity‑rich language; avoid clickbait and ensure facts are structured.
• Track brand mentions in AI answers – Use prompt testing and specialised tools to monitor how often your brand appears in ChatGPT or Perplexity outputs.
• Structure content for AI tools, not just Google – LLMs care about relationships between entities and semantic coherence; build content hubs, use schema markup and organise information clearly.
• Use entity‑rich language – Name people, companies, products and categories directly; this helps AI connect your brand to relevant topic.

The eLearning Industry article on Agentic AI Optimisation (AAIO) offers a similar perspective. It describes AAIO as the practice of optimising content so that AI agents (e.g., AutoGPT or BabyAGI) can autonomously find, analyse and use it. The article emphasises that keywords alone are not enough — metadata, trust signals and structured data matter because AI agents rely on APIs and structured information.

New Metrics and Agency Strategies

With AI summarising content and reducing clicks, agencies must adjust both their strategies and how they measure success. Josh Gross suggests several key actions for agencies:

1. Develop an AI‑use policy. Planetary created a framework that evaluates where AI will be valuable, what data is being used and how the output will be applied. Gross warns against “pushing stuff into tools that you don’t know who’s behind” and treats AI “like a junior‑level team member” whose work needs human review. AI can speed up design inspiration, prototyping and debugging, but human oversight is non‑negotiable.
2. Be transparent with clients. In 2024 some clients prohibited AI; by 2025, many expect agencies to use it. Gross notes that clients now want to know how AI is used and to approve processes before outputs go public. Agencies must set clear expectations for quality control and involve clients in approval workflows.
3. Shift focus from execution to strategy. AI tools make certain tasks faster (asset generation, debugging), but this efficiency raises expectations rather than lowering costs. Clients will “expect more to the output” while rates remain similar. Strategic thinking, creative problem‑solving and innovation become the real differentiators.
4. Measure profitability, relationships and task‑level efficiency. Gross tracks profitability and efficiency (can the team manage projects effectively?), the project‑to‑relationship ratio (moving from one‑off projects to longer‑term engagements) and task‑level efficiency (how quickly the team handles specific tasks). These metrics guide resource allocation and determine when to scale services.
5. Build adaptability into agency DNA. Successful agencies won’t avoid AI or let it do all the work; they will use AI intelligently while doubling down on human capabilities. Gross argues that agencies must evolve their processes, maintain transparency and focus on strategic thinking. Those who fail to adapt risk obsolescence as traditional SEO tactics lose impact.

The Data Behind the Shift

Research and statistics show why AI‑focused SEO can’t be ignored:

• Adoption is exploding. In 2023 about 3 million American adults claimed to have used AI for search; this number is projected to grow to 90 million by 2027. The global AI content‑marketing market, valued at $2.4 billion in 2023, is expected to reach $17.6 billion by 2033 at a compound annual growth rate of 25.68 %.
• Generative AI and marketing are booming. A Digital Marketing Institute survey reports that 92 % of businesses plan to invest in generative AI over the next three years. The AI marketing market could reach $217.33 billion by 2034.
• AI is already embedded in marketing workflows. More than half of marketing teams use AI to optimise content, and many use it for content creation (50 %), brainstorming (45 %), automating repetitive tasks (43 %) and analysing data (41 %). About 73 % say AI plays a key role in creating personalised customer experiences.
• Challenges and skills gaps remain. 31 % of marketers worry about AI accuracy, and 39 % avoid generative AI because they don’t know how to use it safely. Forty‑three percent feel their employers don’t provide adequate training, and many expect performance expectations and tools to change.
• AI won’t eliminate jobs but will change them. Gartner reports that 75 % of companies using AI for marketing plan to shift workers to more strategic activities. A majority of marketing professionals (69 %) feel hopeful about AI’s impact on their jobs.
• AI Overviews are going mainstream. Google’s own blog notes that AI Overviews have been used billions of times through Search Labs and will roll out to all U.S. users, reaching over a billion people by the end of 2025. Google reports that AI Overviews lead to more diverse website visits, and the links in an AI Overview get more clicks than traditional listings. Users will soon be able to adjust AI Overviews (simplify or expand) and ask complex questions that the Gemini model answers in one go.

Choosing and Working With AI SEO Agencies

The How the F*ck article also profiles the leading AI SEO agencies and explains what to look for when hiring one. Key criteria include:

1. Humility and experimentation. No one has all the answers in this emerging field. Choose agencies that test and adapt their methods, and have a playbook that evolves with new findings.
2. Culture of testing. They should be running experiments on LLM behavior and measuring how brand prompts perform inside AI tools.
3. Strong SEO foundations. The best AI SEO agencies are already great at traditional SEO and content writing; the mentality for ranking in LLMs is similar.
4. Data‑led and outcome‑focused. Agencies like Spicy Margarita prioritise bottom‑of‑funnel AI SEO that captures high‑intent queries and ties rankings directly to revenue. They structure content for AI parsing, track brand mentions in AI tools and iterate based on LLM output.

Conclusion: Adaptation Over Hype

The evolution of search in 2025 isn’t about abandoning SEO; it’s about reimagining it for a world where AI tools answer questions and curate information. Agencies that cling to keyword‑stuffing and outdated metrics will struggle to show value as AI overviews and conversational assistants take centre stage. Success will come from:

• Combining technical excellence with clear, structured content and strong brand signals so AI can understand, cite and recommend your site.
• Embracing AI cautiously and thoughtfully, with policies that consider data provenance, privacy and human oversigh.
• Prioritising strategy, creativity and problem‑solving, because efficiency from AI raises the bar for output.
• Educating clients about AI usage and setting expectations around quality and approval processes.
• Watching new metrics, such as AI search visibility, brand mentions and task‑level efficiency.

AI isn’t killing SEO — it’s raising the bar. Agencies that adapt their workflows, rethink their content structures and lean into strategic partnerships will thrive in the AI‑driven search landscape. Those who delay risk getting buried beneath a wave of AI‑generated results.