AI-powered search has fundamentally changed how your customers find content:
- Google AI Overviews appear in 50% of searches (doubled since August 2024)
- AI referral traffic grew 527% year-over-year
- ChatGPT processes 2.5 billion queries daily from 700 million users
- 72% of B2B buyers encounter AI during their research journey
Companies optimising for AI search see 40% visibility improvements and 27% higher conversion rates. Yet only 19% of marketers have a strategy.
AI traffic converts at 3.16x higher rates in insurance, generates 80% more revenue per visit in travel and shows 68% longer engagement times universally. Success is about quality over volume.
-
Written content on site (blogs, white papers, landing pages)
How AI views your written content
AI systems analyse semantic relationships and context, not just keywords. Traditional keyword stuffing performs worse than baseline, while semantic quality optimisation generates up to 40% visibility improvements.
Structure & clarity
- Use simple, declarative sentences in short paragraphs
- Place direct answers in the first 160 characters of content
- Use question-based headers: “What is X?” not “Introduction to X”
- Break content into bullet points and numbered lists
Schema markup
- Add FAQ, Article, Organisation and Product schema
Authority & trust
- Cite credible sources
- Include author bios and company information
- Add FAQ sections addressing common customer questions
- Link to external authoritative sources
Content depth
- Aim for 2,000-3,000 words for comprehensive topics
- Content over 3,000 words generates 3x more traffic
- 88.1% of AI queries are informational
File formats
- Keep gated content (like PDFs) mirrored with an HTML summary page / overview synopsis web page — AI tools struggle to crawl PDFs behind forms.
-
Publishing & technical setup
Unlike Google’s sophisticated crawler, most AI bots make simple requests with tight timeouts. Content that loads slowly or requires JavaScript execution won’t be seen.
Crawlability
- Don’t block beneficial AI crawlers in robots.txt
- Allow: ChatGPT-User, PerplexityBot, Googlebot
- Consider blocking training crawlers: GPTBot, CCBot (your choice)
- Blocking Google-Extended does not affect Google Search rankings
- Ensure clean sitemaps
Content freshness
- Update key pages regularly – AI prefers recent sources
- 85% of Google AI citations published within last 2 years
- Add current year to titles: “2025 Guide to…” boosts citation rates
- New content boost: 2-3 days at peak, then decays to baseline
Page speed
- AI crawlers timeout at 1-5 seconds – fast loading is critical
- Mobile optimisation is essential
Alt text & captions
- Add descriptive alt text to all images (125 characters)
- Instead of “infographic,” write: “2025 B2B SaaS market growth by company size and region”
- Video transcripts are essential (more below)
-
Designed content (Infographics, visual assets)
AI multimodal models can extract text from images with varying degrees of accuracy. For best practice, pair visual content with text equivalents to ensure AI can access and cite your information.
Provide text equivalents
- Always pair infographics with written breakdown beneath
- Publish as HTML article, not just PNG/PDF
- Make data from visuals available as text
Detailed alt text & captions
- Be specific: “Infographic showing UK graduate employment trends 2025, broken into 3 categories: tech, finance, retail”
- Include names, numbers, locations – AI recognises named entities
Descriptive File Naming
- Use: “uk-graduate-employment-trends-2025.png”
- Avoid: “final_version2.png”
Example improvements:
- ❌ “Market Overview 2025”
- ✅ “B2B SaaS Market Growth 2025: Annual Recurring Revenue Trends Across US and UK Markets by Company Size”
-
Video (animated & real-shot)
AI employs multi-stage pipelines combining audio transcription, visual analysis, and metadata extraction. Transcripts are the primary way AI “reads” video content.
Transcripts & captions
- Upload full, accurate transcripts (99% accuracy minimum)
- Use crawlable formats: WebVTT or SRT
- AI prioritises text transcripts over video analysis
- Publish transcripts as accompanying blog posts
Structured video data
- Use VideoObject schema with title, description, duration, thumbnail
- Include comprehensive descriptions with key topics
- Add timestamps for navigation
Summaries & key takeaways
- Publish “Highlights” or “What you’ll learn” section in plain text
- Provide quick answers (first 40-60 words) plus detailed explanations
Chapters & timestamps
- Break long videos into chapters with descriptive names
- AI pulls answers from specific moments using metadata
- YouTube chapters improve both UX and AI extraction
Descriptive titles
- ❌ “Product Demo Video”
- ✅ “Complete Salesforce Integration Demo 2025: Step-by-Step Setup Guide for Enterprise CRM Migration (12:34)”
-
General AI-search optimisation tips
Only 11% overlap between ChatGPT and Perplexity citations – different platforms use different sources.
Platform preferences:
| Platform | Top citation sources | What this means |
| ChatGPT | Wikipedia (47.9%)
Encyclopedic sources |
Favours factual, comprehensive, well-structured content |
| Google AI Overviews | Reddit (21%)
YouTube (18.8%) Quora (14.3%) LinkedIn (13%) |
Most diverse – values community discussions, video content, and professional platforms |
| Perplexity | Reddit (46.7%)
YouTube (13.9%) Gartner (7%) |
Heavily favours community discussions and authoritative research |
Success on one platform provides minimal benefit on others – you need multi-platform presence.
Answer questions directly
- Write in Q&A style – AI retrieves direct answers
- Use conversational language (23-word queries vs. 4-word traditional)
- Front-load value propositions with names, numbers, benefits
Optimise for Natural Language
- Write how people speak and ask questions
- Focus on “how,” “what,” “why” questions
Ensure accessibility
- Every format (visual, video, PDF) needs text equivalent
- Text versions make insights discoverable and citable
- Critical for JavaScript-limited AI crawlers
Build trust & authority
- Cite credible sources consistently
- Include author expertise and credentials
Think “chunking”
- Break content into modular, standalone sections
- One idea per paragraph with clear H2/H3 subheadings
- Each section should make sense when extracted alone
- AI uses “chunk” retrieval, not full-page processing
Multi-platform presence for B2B
- Participate authentically on Reddit (technical discussions)
- Build thought leadership on LinkedIn
- Create educational YouTube content
- These drive AI citations independent of your website
Implementation roadmap
Phase 1: Quick wins
Focus on highest-impact, lowest-effort actions:
- Set up tracking – Add GA4 tracking for AI traffic sources (ChatGPT, Perplexity, Gemini, Claude)
- Audit robots.txt – Ensure beneficial crawlers can access content
- Add basic schema – Organisation and Website schema on homepage with links to LinkedIn, Wikipedia
- Optimise top pages – Rewrite top 10 pages with answer-first structure (key message in first 160 characters)
- Claim profiles – Google Business Profile for all locations (Google AI Mode prioritises local info)
Phase 2: Build infrastructure
- Expand schema – FAQ, Article, HowTo, Product schemas across key templates
- Launch research – Original data/statistics programme
- Platform presence – Establish authentic presence on Reddit, LinkedIn, YouTube
- Voice optimisation – Add conversational keywords and question-based formatting
- Start monitoring – Use AEO tools (Profound, Ahrefs Brand Radar, SE Ranking) to track citations
Phase 3: Long-term strategy
- Multi-format content – Videos with transcripts, infographics with HTML breakdowns, podcasts
- Entity graphs – Structured data connecting your organisation, people, services
- Digital PR – Target high-citation publications (platform-specific)
- Governance – Schema maintenance processes as content evolves
- Integration – Unify AEO with paid, social and PR channels
Budget guidance
Current reality: Google processes 210x more searches than ChatGPT, but AI traffic grows 165x faster than organic search.
2025 recommendation: 70% traditional SEO / 30% AI optimisation
2026 shift: Move towards 50/50 as AI penetration deepens
Understanding the zero-click future
Traditional content marketing measured: traffic volume, rankings, conversions
AI search era demands new metrics:
- 60% of searches end without clicks (rising to 69% globally)
- When Google AI Mode activates: 93% end without clicks
- AI Overviews generate zero clicks 43% of the time
Success = being cited and mentioned, not receiving clicks
Case study: NerdWallet achieved 35% revenue growth despite 20% traffic decrease by focusing on high-quality, AI-visible content
After implementing AI optimisation (90 days):
- 10% of total traffic from LLMs
- 27% of AI traffic converted to sales-qualified leads
- 30% higher time on site for LLM visitors
- Lead quality “significantly better” -buyers arrived ready to purchase
Key takeaways
- Act now while competition is limited – Only 19% of marketers have AI strategies
- Quality beats volume – AI traffic converts at 3.16x higher rates
- Multi-platform is essential – 89% of citations differ between platforms
- Structure matters more than keywords – Semantic clarity and schema drive AI citations
- Think citations, not traffic – 60%+ of searches now end without clicks
- Text equivalents are critical – AI cannot reliably read images or execute JavaScript
- Transcripts unlock video discoverability – Primary way AI “reads” video content
Next steps: Questions for your team
To assess current state:
- What percentage of our traffic comes from AI sources? (Check GA4)
- Can AI crawlers read our site, or do we need server-side rendering?
- Do we have schema markup implemented? Which types?
- Do our videos have full, accurate transcripts?
- Are our infographics paired with HTML text breakdowns?
- Where does our target audience discuss industry topics? (Reddit, LinkedIn, forums)
Ready to optimise your content for AI search? Let’s talk about how we can help you get discovered, cited, and ahead of your competition in the AI search era.