SEO content now serves three distinct audiences — human readers, traditional search engines, and AI models that generate overviews and chatbot responses. Our team has spent the past quarter retooling client strategies around this reality. The shift demands new tracking methods, new research workflows, and a sharper understanding of where brands appear (and where they don’t) inside AI-generated answers. Here’s what we’re acting on right now.

Key Takeaways

• AI Overviews are expanding across Google SERPs, and tracking citation presence is now a core SEO metric.
• Content must be written for users, crawlers, and large language models simultaneously — a three-audience framework.
• Prompt research is an emerging discipline that reveals where a brand competes inside AI responses and where it’s invisible.
• AI visibility gap analysis lets agencies pinpoint the exact prompts and sources competitors get cited in.
• A new role — the Content SEO Manager — is consolidating content strategy, technical SEO, and AI expertise into a single position.

Tracking AI Overview Citations Is Now a Standard Client Deliverable

Google’s AI Overviews are no longer experimental. They appear across an increasing share of commercial and informational queries, pulling cited sources directly into the SERP. If our clients’ pages aren’t cited, traffic erodes — regardless of traditional ranking position.

We’ve integrated AI Overview monitoring into every active campaign, a workflow supported by Semrush’s AI Overview tracking capabilities. The practical output: weekly reports showing which client URLs appear in AI-generated summaries, which queries trigger overviews, and where citation share is rising or falling. This data feeds directly into content prioritisation decisions.

Writing for Three Audiences at Once

The old two-audience model (write for people, optimise for bots) is outdated. AI models now parse, summarise, and cite web content independently of Google’s traditional index. That means structure, factual density, and source authority all carry more weight than before.

Our content team follows a five-step production workflow aligned with Semrush’s complete guide to SEO content for search and AI. In practice, this means:

• Leading every page with a clear, direct answer to the target query.
• Using structured data and consistent heading hierarchies so LLMs can extract facts cleanly.
• Including original data, named sources, and specific figures — the signals AI models weigh when selecting citations.

Pages built this way rank in traditional results and get pulled into AI-generated answers. That dual visibility is the baseline standard we hold every deliverable to.

Prompt Research: Finding Where Your Brand Does and Doesn’t Appear

Keyword research alone no longer captures the full picture. Users now ask AI chatbots open-ended questions — prompts — and the brands that get named in those responses gain a distinct advantage.

We run structured prompt research for every client using a repeatable process detailed in Semrush’s guide to prompt research for AI SEO. The method maps a brand’s presence across common conversational queries in its category, identifies gaps, and feeds those gaps back into the editorial calendar. It’s straightforward competitive intelligence applied to a new surface.

Closing AI Visibility Gaps Before Competitors Widen Them

Knowing where competitors get cited — and you don’t — is actionable intelligence. We use gap analysis to surface the specific prompts and third-party sources driving competitor citations inside AI responses, following the methodology outlined in Semrush’s AI visibility gap workflow.

Once gaps are identified, we prioritise content creation or optimisation around those prompts. The goal is simple: close the gap before competitors consolidate their citation advantage.

The Content SEO Manager Role Is Reshaping Agency Hiring

A dedicated study analysing over 1,000 job listings confirms what we’ve observed internally: the Content SEO Manager role is becoming the central hub for content strategy, technical SEO, and AI fluency. The findings, published in Semrush’s Content SEO Manager study, show employers increasingly expect a single hire to bridge editorial quality and search performance — including AI citation strategy.

For our agency, this validates the cross-functional skill sets we’ve been building across the team throughout 2025. Every strategist working on client accounts now operates with content, technical, and AI visibility responsibilities rolled into one brief.

generative engine Optimisation is not a future consideration — it is a present operational requirement. Agencies that treat AI citation tracking, prompt research, and three-audience content production as standard practice will hold a measurable advantage heading into 2026. We’re building every client engagement around these disciplines now.

Frequently Asked Questions

What is Generative Engine Optimisation (GEO)?

GEO is the practice of optimising web content so it gets cited in AI-generated responses from tools like Google’s AI Overviews and AI chatbots. It extends traditional SEO by targeting how large language models select and reference sources.

How do web designers ensure pages get cited in AI Overviews?

Pages need clear, direct answers near the top, structured heading hierarchies, and original data or named sources that AI models can extract. Combining these elements with strong technical SEO gives pages the best chance of appearing in AI-generated summaries.

Why does prompt research matter for SEO in 2025?

Users increasingly ask AI tools open-ended questions rather than typing short keywords. Prompt research identifies which conversational queries mention your brand and which don’t, so you can create targeted content to fill those gaps.

What is a Content SEO Manager and why is the role growing?

A Content SEO Manager combines editorial strategy, technical SEO, and AI visibility skills into one position. The role is growing because businesses need a single point of accountability for content that ranks in traditional search and gets cited by AI models.