Content marketing is king — but what do you write when AI writes everything?
Content marketing still works — but now anyone can generate infinite mediocre content in seconds, so the bar moved. The question isn't just 'what do I write,' it's 'what can I write that AI can't, and that AI will cite.' Here's the answer.
Content marketing works. That was true when we first wrote this, and it's still true. The old hard part was simply writing — for most of us, producing something interesting was laborious and slow. Well, that problem got solved, and not in the way anyone hoped: now anyone can generate infinite, competent-sounding content in seconds. Which means the question changed. It's no longer "what do I write," it's "what can I write that AI can't — and that AI will actually cite?"
Why "just publish something" stopped being enough
The 2011 advice was to follow your interest and hit publish, because getting decent content out beat polishing forever. That instinct still matters — done beats perfect. But the floor collapsed: the web is now flooded with AI-generated filler, and Google's helpful-content and spam systems actively demote generic, expertise-free content. Publishing another competent-but-hollow post doesn't move the needle anymore; it just adds to the noise everyone's trying to filter out.
What to write now: the stuff AI can't fake
The differentiator is exactly what generative models don't have — lived experience and proprietary knowledge. Write the things only you can:
- First-hand experience. What you actually did, what happened, what you'd do differently. "We ran this campaign and here's what broke" is something no model can generate, and it's what Google's E-E-A-T (the extra E is for Experience) rewards.
- Original data. Your numbers, benchmarks, a survey of your customers. Original stats are the single most-cited, most-linked kind of content — by journalists and by AI answer engines alike.
- A real point of view. An opinion you'll defend, a contrarian take, a framework you developed. AI averages the internet; it can't have a spine. You can.
Write to be cited, not just ranked
There's a new reason this matters. People increasingly get answers from AI assistants — ChatGPT, Claude, Gemini, Perplexity — that assemble responses by pulling from authoritative sources. Generic content never gets pulled; distinctive, expert, well-structured content does. So the modern content goal isn't only to rank in Google, it's to become a source the answer engines quote. That's AEO/GEO, and it rewards exactly the genuine-expertise content AI can't mass-produce.
The kernel that survives
The old advice wasn't wrong, just incomplete: follow what genuinely interests you and get it out there — but make sure it carries something only you have. In an AI-flooded world, your unfair advantage is your actual experience and point of view. Write from that, structure it clearly, and publish. That's the foundation of how we approach SEO, AEO & GEO — content that earns rankings and citations because there's something real behind it.
What should I write about for content marketing now that AI exists?
Write the things AI can't generate: first-hand experience (what you actually did and learned), original data (your own numbers, benchmarks or surveys), and a genuine point of view. Generic, competent-sounding content is now infinite and worthless; distinctive content grounded in real expertise is what ranks, earns links, and gets cited by AI answer engines.
Is AI-generated content bad for SEO?
Generic AI content is risky. Google's helpful-content and spam systems demote mass-produced, expertise-free content regardless of how it was made. AI can be a useful drafting and research tool, but content that's purely AI-generated filler with no real experience or original insight tends to get filtered out. The differentiator is genuine value, not the tool used to write it.
What does it mean to write content that AI will cite?
People increasingly get answers from AI assistants that assemble responses from authoritative sources. To be one of those sources, your content needs to be distinctive and genuinely expert — original data, clear answers, real experience — not generic filler the model can already produce itself. Optimizing for these citations is what AEO (answer engine optimization) and GEO (generative engine optimization) are about.