Devstars
Blog
Date: 18/11/2025
Stuart WatkinsGoogle’s AI now decides whether your business gets mentioned. Here’s how to make sure you’re in the conversation.
AI search is growing fast, but it hasn’t replaced traditional search yet. The data tells a clear story.
Google still processes 16.4 billion searches daily compared to ChatGPT’s 1 billion queries. Nearly 80% of Americans still prefer traditional search engines. Millennials and Gen Z are adopting AI tools faster, but we’re in the early adoption phase.
Here’s the thing, though. AI Overviews now appear in roughly 60% of Google searches. When they do, only 8% of people click through to websites. But those who do click convert 4.4 times better than traditional visitors. Meanwhile, 59% of all searches now end without a click at all.
So the question isn’t whether optimising for AI search matters. It’s whether your website is structured to be cited when it does.
We’ve spent 6 years implementing schema markup and structured data for clients. What was once a nice-to-have is now the difference between visibility and invisibility.
Here’s the encouraging truth. Generative Engine Optimisation, or GEO, isn’t a completely new discipline. AI platforms typically pull from top-ranking search results and trusted sources to compose their answers. Strong traditional SEO remains the foundation.
Google’s own advice, stated at the November 2025 Search Central event, was straightforward: “To get your content to appear in AI Overview, simply use normal SEO practices.”
But “normal SEO practices” now means being more rigorous about the fundamentals. The businesses winning in AI search aren’t doing anything revolutionary. They’re doing the basics properly, which most websites still don’t.
What’s changed is the reward system. Traditional SEO earned you a blue link. Optimising for AI search earns you a citation, a direct mention in the answer your potential customer reads. And the data shows that 48% of AI Overview sources come from pages outside the traditional top 10 results. Well-structured content that wouldn’t have dominated page one can now get cited directly in the answer.
Google’s enterprise documentation recently confirmed what we’ve suspected: AI search uses multiple signals to decide which content to cite. Semantic understanding, keyword relevance, click behaviour, freshness, and crucially, how well-structured your content is.
The documentation reveals a “grounding” system that checks whether content can actually support claims. Vague statements get filtered out. Specific, verifiable information gets cited.
What that means practically is that AI systems are looking for content they can confidently quote. Here’s what works.
AI systems extract from the first 50-70 words. If your opening is throat-clearing about your company history, you’ve wasted the prime real estate. Lead with the answer.
“How much does a website redesign cost?” as a heading, followed by specific pricing, is citable. “Our Pricing Philosophy” is not. Frame headings as natural, complete questions that mirror actual search queries.
“Increased conversions by 47% over 6 months” can be cited. “Significantly improved results” cannot. Numbers, dates, and specifics give AI systems confidence. This is where case studies and real data become your competitive edge.
AI platforms don’t just match keywords. They understand concepts. Use varied related terms rather than keyword repetition. Cover topics thoroughly within individual sections so AI can cite a complete answer without stitching together fragments from multiple pages.
AI systems cite content that fully addresses a query within a single section. Partial answers that require clicking through multiple pages don’t get selected. Deep pages receive 82% of citations versus homepages, because AI favours specific, detailed content over general landing pages.
Quick example that might help: when the Royal Academy of Music needed their events to stand out in search, we implemented Event schema across their listings. Concerts and performances appeared with dates, venues, and booking links directly in search results, driving ticket sales from search visibility alone. For SlimFast, we structured recipe content with Recipe schema, including nutritional data, cooking times, and ingredients. The result was a diet brand’s website outranking BBC Good Food for competitive recipe searches. Today, that same schema gives AI systems the structured data they need to cite SlimFast as a credible source.
The technical side of optimising for AI search isn’t glamorous, but it’s non-negotiable. AI agents literally abandon broken pages. If your site has errors, slow speeds, redirect loops, or missing schema, you’re invisible before you even start.
Schema markup is code that tells search engines precisely what your content represents. A recipe. An event. A service. A review. Without it, AI systems have to guess. With it, they know.
There are over 800 schema types available. Most websites use two or three, often incorrectly. We regularly audit sites that have schema markup that doesn’t match the page content, missing schema on pages that desperately need it, outdated implementations, or schema that’s technically present but configured wrong.
Priority schema types for most businesses:
Controlled experiments published by Search Engine Land in 2025 found that pages with well-implemented schema were the only ones to appear in AI Overviews in their test. That’s a pretty strong signal.
Remember, no secret AI tags exist. Use established structured data types and implement them correctly.
Check your robots.txt files. Don’t block OpenAI’s GPTBot, Google’s crawlers, or Microsoft’s Bingbot. If you’re accidentally telling AI platforms not to read your content, you’ve ruled yourself out.
Ensure key content lives in HTML text, not hidden behind JavaScript. Include descriptive alt text for images and transcripts for videos. Maintain a clean site hierarchy and fast loading speeds.
What we’re seeing across our client base is that messy site structure confuses users and AI alike. If your team has been adding pages for years without a clear plan, you’ve got a filing cabinet disaster. Clean, logical categorisation helps users find what they need, helps Google understand your priorities, and helps AI platforms parse your content correctly.
Over 60% of B2B research happens on mobile now. Poor mobile experience means lost opportunities. Review your site on actual mobile devices, not just responsive preview. Fix navigation, form completion, and page speed issues. Test the entire customer journey.
AI platforms don’t just assess individual pages. They evaluate whether your brand is a credible source across the wider web.
Demonstrate expertise with author bios, credentials, and detailed case studies. Every team member who creates content should have a proper author profile with relevant qualifications. This isn’t vanity. It’s how AI systems assess whether your content is trustworthy enough to cite.
AI Overviews mention an average of 5 sources per query, and multiple pieces from the same domain can be featured. Building deep topical authority across your niche means you could appear more than once in a single AI response.
The old model was backlinks. Links still matter, but context around your brand mentions matters more now. AI doesn’t need a hyperlink to understand authority. It reads the surrounding text to understand how other sites describe you. Your positioning statements, USPs, and the way industry publications reference your work, that’s what AI reads.
What still works well is digital PR and journalist commentary, charity partnerships and sponsorships, industry-specific directories (not generic spam ones), and brand mentions with context. What doesn’t work is buying 500 cheap links from link farms. Always was worthless, definitely is now.
AI systems frequently cite user-generated content platforms. Quora is the most commonly cited source in AI Overviews, followed by Reddit, YouTube, LinkedIn, and major publishers.
For any business serious about optimising for AI search, establishing presence on these platforms matters as much as on-site work. Answer industry questions on Quora. Participate in relevant Reddit discussions. Publish LinkedIn Articles demonstrating your expertise. Create YouTube content addressing your most-asked questions.
YouTube deserves particular attention. It processes 3 billion daily searches now. People search YouTube before Google for how-to content, product reviews, and solutions. If you’re not there, you’re invisible to that entire audience. Create 6-12 pillar videos addressing your most common questions. Optimise titles and descriptions for search intent, not just views.
Static websites that don’t evolve are haemorrhaging traffic to competitors who publish fresh, relevant content regularly. Citation volatility in AI Overviews is high, with roughly 70% of cited pages changing within 2-3 months. This isn’t a one-off project. It’s an ongoing commitment.
Content needs 3-6 months to establish trust signals before achieving optimal AI citation rates. So the content you publish today is your visibility in late 2026. Start now.
Competitors using AI properly are producing 5-10 times more content at similar quality. Pure human-only approaches can’t compete on volume.
What works is an expert-AI partnership model. Human provides the expertise and strategic direction. AI assists with structure and scale. Human reviews for quality and those E-E-A-T signals that AI alone can’t provide. We call this the 10-80-10 approach: 10% human strategic input, 80% AI execution, 10% human refinement and quality control.
But here’s an important caveat. One-prompt blog posts, the kind where you type “write me a blog on X” and publish whatever comes back, create worthless content that won’t rank or convert. Use AI as a conversational tool. Edit, refine, personalise. Add your actual expertise and experience.
One e-commerce client saw a 280% organic revenue increase using seasonal keyword campaigns. They created content guides for each season published just before peak demand. One piece hit Google Discover and drove a 940% traffic spike during peak season.
Map your keywords to demand cycles. Create timely, relevant content ahead of your busy periods.
Traditional click tracking isn’t fully effective for AI search since users often get answers without clicking through. Marketing attribution is fundamentally broken in 2026, with AI tools, privacy changes, and multi-device journeys making it nearly impossible to track where every lead originates.
Here’s what to do instead.
Monitor AI visibility directly. Tools like Semrush AI Toolkit, SE Ranking AI Tracker, and free options like HubSpot AI Search Grader and Advanced Web Ranking’s AI Overview Tool can track where your content appears across AI platforms.
Track brand mentions across AI platforms. When someone asks ChatGPT, Claude, or Perplexity about your industry, does your brand come up? This is your new visibility metric.
Ask every new enquiry “How did you find us?” Build a simple qualitative tracking system. Focus budget on channels that consistently appear in real customer stories, not algorithmic fiction.
Prioritise conversions over traffic. Traffic is a progress marker. Conversions are the goal. If you’re losing informational traffic to AI Overviews but maintaining commercial traffic and conversions, you’re winning.
Run conversion rate audits. With declining click-through rates, squeezing more value from existing visitors is more cost-effective than chasing more traffic. Fix the obvious wins: unclear value propositions, weak calls-to-action, mobile experience issues, slow load times. A 20-30% improvement in conversion rate from the same traffic is easier and cheaper than buying 30% more visitors.
Optimising for AI search isn’t about learning entirely new tactics. It’s about applying SEO fundamentals more rigorously, with particular attention to content structure, schema implementation, technical accessibility, and comprehensive topic coverage.
AI search isn’t replacing traditional SEO. It’s adding a layer that rewards structure, specificity, and proper technical implementation. The websites that prepare now will capture visibility whilst competitors wonder why their traffic dropped.
The businesses that will thrive in 2026 are those that started adapting to the AI-first search landscape in 2025, not those still optimising for how search worked in 2023.
Schema markup has moved from competitive advantage to baseline requirement. If your website isn’t speaking the language AI systems understand, you’re not in the conversation.
Want to know where your website stands? We offer AI readiness audits that assess your current structure, schema implementation, and content against what AI search rewards. Get in touch and we’ll show you exactly what needs attention.
Stuart Watkins is Founder of Devstars, a Jersey-based digital agency that’s been implementing schema markup and structured data since 2019. When AI search started rewarding structure, his clients were already prepared.
Currently scheduling strategic partnerships for Q1-Q2 2026. Limited spaces remain.
Get a free technical consultation and project roadmap. We’ll assess your requirements and provide transparent pricing for your growth-stage development needs.
Call: +44 020 8898 3993
