SEO for Humans, not LLMs: Recovering Traffic in the SGE Era

In early 2024, I audited the content strategy for a $30M ARR RevOps software company. They had 140 blog posts. 90% of their traffic was going to five articles — all of them definitional pieces like "What is Revenue Operations?" and "RevOps vs SalesOps: What's the Difference?"

By mid-2025, those five articles had lost 60% of their traffic. Not because of algorithm changes. Because GPT-4, Perplexity, and Google's own AI Overviews were answering those questions directly in the search results, faster and more thoroughly than any article could.

The SEO strategy that built most B2B content libraries is the exact strategy that's dying fastest in the AI search era. And the companies that haven't adapted yet are about to find out the hard way.

What AI Search Actually Changed

Search Generative Experience (SGE) and AI-powered answer engines don't just steal traffic from definitional content. They've fundamentally changed the intent of who reaches your content organically.

The buyers who are now clicking through to long-form content are not looking for definitions. They already know what RevOps is. They're looking for something an AI can't synthesise: your specific experience, your controversial take, your proprietary data, and your demonstrated judgment. They want the thing that only a practitioner who has actually done this work can provide.

This is a gift if you understand it. Your traffic will get smaller. Your readers will be better.

The Information Gain Framework

Google's Helpful Content System uses a concept their quality raters call "Information Gain" — the degree to which a piece of content adds something that doesn't already exist in other sources. In the AI search era, this concept becomes the primary filter for organic visibility.

High Information Gain content has specific properties:

It contains proprietary data. Benchmark data from your own customer base. A/B test results from campaigns you ran. Win/loss analysis patterns from your pipeline. An LLM can't hallucinate this because it's yours alone.

It reflects lived experience. "In the 40 audits I've run this year..." tells a reader something an AI can't fabricate. First-person practitioner narrative carries E-E-A-T weight that generic how-to content never will.

It takes an uncomfortable position. "Stage-based forecasting is causing your CFO to mistrust your pipeline reports" is an opinion a reader has to engage with. It creates what marketers call a "pattern interrupt." Compare that to "Pipeline forecasting best practices" — which an AI can, and will, render perfectly.

It creates a specific, citable framework. The more original and named your thinking, the more likely it is to be referenced, shared in communities, cited by other writers, and used as a source by AI systems that need to cite something credible.

What to Build Instead

Original research and benchmarks. Survey your customers. Analyse your own anonymised data. Publish findings with your methodology visible. This is the highest-value SEO content in 2026 because it's unfakeable.

Case studies with specific numbers. Not "we helped a company improve their pipeline." Rather: "How we reduced lead response time from 4.2 hours to 11 minutes across 3 enterprise sales teams." Specificity is the signal of authenticity that AI can't replicate.

Perspective pieces that challenge consensus. The best-performing articles in our network in 2025 were not the most comprehensive. They were the most polarising. A piece titled "Your ABM programme is probably just expensive outbound" generates more shares, more comments, and more community discussion than "The Complete Guide to ABM" ever will.

The community integration. Content that gets discussed in communities gets distributed in dark social channels, which drives branded search, which signals relevance to both Google and AI training pipelines. Write for practitioners, publish in community spaces, and let the distribution compound.

The Honest Challenge

This shift requires a fundamentally different content production model. You can't produce 30 SEO-optimised articles per month and call it a strategy. You need to produce fewer, deeper, more original pieces — and you need to source them from people who have actually done the work, not from writers briefed with keyword lists.

For most teams, this means building a contributors programme that brings genuine practitioners into the content process. The authority in the article has to come from the person who wrote it, not from the topic.

The AI search era doesn't make content marketing obsolete. It makes mediocre content marketing obsolete. That's probably fine with you.