This post was originally published on my portfolio.
For 25 years, getting found online meant one thing: rank on Google. Build backlinks, stuff keywords, chase PageRank. The rules were clear even if the game was dirty.
That era is ending. Not slowly — fast.
In 2024, Google launched AI Overviews. ChatGPT crossed 100 million weekly users. Perplexity grew 10x. People stopped clicking blue links and started asking questions and getting answers. The search engine is becoming an answer engine — and the optimization playbook has changed completely.
Here's what the data shows, what changed, and exactly what you need to do about it.
The numbers that changed everything
Google search click-through rates are collapsing
| Year | Average organic CTR (position 1) | Zero-click searches |
|---|---|---|
| 2019 | 28.5% | 49% |
| 2021 | 25.6% | 57% |
| 2023 | 22.1% | 65% |
| 2024 | 18.4% | 72% |
| 2025 | ~14% (est.) | ~79% (est.) |
Sources: SparkToro (2024), Semrush Search Behaviour Report (2024), Rand Fishkin / SparkToro Zero-Click Study
Even if you rank #1 on Google, nearly 4 in 5 searches now end without a click. Google answers the question itself — via featured snippets, AI Overviews, Knowledge Panels, and People Also Ask boxes.
LLM usage is growing at a rate search never did
| Platform | Monthly active users (2024) | YoY growth |
|---|---|---|
| Google Search | 8.5 billion queries/day | +2% |
| ChatGPT | 100M+ weekly users | +300% |
| Perplexity | 10M+ monthly users | +900% |
| Microsoft Copilot (Bing AI) | 5B+ chats (cumulative) | N/A |
| Claude (Anthropic) | 30M+ monthly users | +400% |
| Gemini (Google) | Integrated into 1B+ devices | N/A |
Sources: OpenAI blog (2024), Perplexity investor deck (2024), Microsoft Build keynote (2024), Anthropic usage reports
Traditional SEO vs LLM Optimisation — side by side
| Factor | Traditional SEO (Google) | LLM Optimisation (AEO) |
|---|---|---|
| Goal | Rank on page 1 | Be cited as a source |
| Signal | Backlinks, domain authority | Semantic clarity, structured data |
| Content format | Keyword-dense, long-form | Clear, factual, well-structured |
| Key file |
sitemap.xml, robots.txt
|
llms.txt, robots.txt with AI crawlers allowed |
| Metadata | Title tags, meta description | JSON-LD schema, og: tags |
| Discovery | Googlebot crawl | GPTBot, ClaudeBot, PerplexityBot crawl |
| Trust signal | PageRank, backlinks | Named entity recognition, structured facts |
| Measurement | Rankings, organic traffic | Citations in AI answers, brand mentions |
| Optimise for | Algorithm | Understanding |
What changed in each era
Era 1: PageRank (1998–2010)
Google's original insight was that links = votes. More links from authoritative sites = higher rank. SEO meant building backlinks, often artificially.
Optimise for: Quantity of inbound links
Era 2: Content & Keywords (2010–2018)
Panda (2011) and Penguin (2012) killed link spam. Content quality started mattering. Keyword research became a discipline. Long-form content won.
Optimise for: Keyword relevance + content depth
Era 3: E-E-A-T & Structured Data (2018–2023)
Google's Quality Rater Guidelines introduced Experience, Expertise, Authoritativeness, Trustworthiness. Schema markup (JSON-LD) let sites communicate structured facts directly to Google.
Optimise for: Author credibility + structured data
Era 4: Answer Engines & LLMs (2024–present)
AI Overviews, ChatGPT, Perplexity, Claude. The answer is generated, not linked. Your content either gets synthesised into the answer or it doesn't exist.
Optimise for: Being understood and cited by AI
The new technical checklist
What still matters from traditional SEO
- ✅
sitemap.xml— LLM crawlers use it too - ✅ Fast page load — still a ranking signal
- ✅ Mobile-friendly — non-negotiable
- ✅ HTTPS — baseline trust signal
- ✅ Clean URL structure
What's new for LLM optimisation
- ✅
llms.txt— new standard at llmstxt.org - ✅
robots.txt— explicitly allow GPTBot, ClaudeBot, Google-Extended, PerplexityBot - ✅ JSON-LD structured data —
Person,Article,WebSiteschemas - ✅ IndexNow — push URLs to search engines instantly on publish
- ✅ Named entity clarity — state who you are, what you do, where you are, explicitly
- ✅ Factual, citable content — LLMs cite sources that state facts clearly
What no longer works
- ❌ Keyword stuffing
- ❌ Thin content optimised for one keyword
- ❌ Blocking AI crawlers
- ❌ Optimising only for click-through
The citation economy
Traditional SEO was a ranking economy — position 1 gets 28% of clicks, position 10 gets 2%.
LLM optimisation is a citation economy — either your content is used to generate the answer, or it isn't. There's no position 2.
The way to win citations:
- State facts clearly — LLMs extract factual sentences
- Use structured data — JSON-LD tells crawlers exactly what type of thing you are
- Be consistent across the web — your LinkedIn, GitHub, and website should all describe you the same way
- Publish original insights — LLMs prefer primary sources over aggregators
- Update content — stale content gets deprioritised
What to do this week
- Add
llms.txtto your root - Update
robots.txtto allow AI crawlers - Add JSON-LD
Personschema to your homepage - Submit
sitemap.xmlto Google Search Console and Bing Webmaster Tools - Set up IndexNow for instant indexing on publish
Building your own LLM-optimised presence? Connect with me on LinkedIn or read more on my portfolio.
Top comments (0)