I’m going to say the quiet part out loud: 99% of the content your agency is producing is expensive, ineffective, digital landfill.
We stopped writing blog posts a year ago. We had to. The model was broken. We were creating these beautifully crafted, 2,000-word monoliths for our clients, agonising over tone of voice and elegant prose. We’d hit publish, get a little spike of traffic, and then… nothing. A slow, quiet decay into the depths of page 14 of Google Search.
The ROI wasn’t there. The effort-to-impact ratio was appallingly low. We were winning awards for creativity while losing the war for attention and revenue. That’s a founder’s nightmare. It’s the moment you realise you’re selling a beautifully packaged lie.
The inflection point came when I looked at the outputs from one of our internal agentic AI projects, Competable. It was spitting out thousands of lines of structured, semantically rich competitive intelligence data in minutes. Data that was more useful, more actionable, and frankly, more valuable than the article our content team had spent three weeks writing. My code was outperforming my copywriters. That’s a problem. Or, as I saw it, an opportunity.
We killed the traditional blog post. We stopped writing for humans first. We started writing for machines.
To be clear, we didn't stop writing. We stopped writing blog posts. We replaced them with a new methodology I developed: Google-as-an-Endpoint Optimisation, or GEO.
This isn't just a rebrand of SEO. It’s a fundamental mindset shift. SEO is a game of ranking. GEO is a game of direct information transfer. It treats Google not as a search engine to be climbed, but as an endpoint to be populated. The goal is no longer to rank #1. The goal is to become the answer.
In a world of Google’s AI Overviews, where the search engine scrapes, synthesises, and presents information directly, the click is becoming a secondary metric. Your content's job is to be so ridiculously well-structured, factually dense, and semantically clear that you become the foundational source for the AI's answer. You aren't trying to get the user to your site; you're trying to get your data into the user's answer.
1. Hyper-Structured Data: Every single piece of content is built on a bedrock of Schema markup, JSON-LD, and other structured data formats. We don’t just write about a product's price; we explicitly tag <"price": "99.99">. We define entities, relationships, and properties with machine-readable precision.
2. Semantic Saturation: We move beyond keywords. We build content around entities and topics. We use natural language processing (NLP) tools to analyse the semantic relationships between concepts, ensuring our content covers a topic with the same breadth and depth as an encyclopedia entry, but with the clarity of a technical manual.
3. Entity Resolution: Google’s knowledge graph is built on entities — people, places, things, concepts. Our primary job in GEO is to create, define, and connect these entities. We want Google to see our client, let's say a UK challenger bank, not as a website with keywords like 'savings account', but as a definitive entity inextricably linked to concepts like 'ethical finance', 'high-interest ISAs', and 'digital banking'. A prime example of a brand failing at this is Tesco. Search for 'compostable coffee pods' – a huge and growing consumer interest. You won't find Tesco's own-brand pods dominating that AI Overview, despite their scale. They haven't established themselves as the entity for that query. They are just another retailer selling a product. They haven't become the source.
Under the GEO model, the article is a delivery mechanism, not the product itself. The product is the data within it.