Programmatic SEO vs. LLM Discovery: A Post-Update Reality Check

The May and June 2026 Google updates completely flattened my programmatic SEO experiment. After five months of steady growth and 50 daily impressions, the site traffic went from a peak to absolute zero in just three weeks. It turns out the old way of spinning up hundreds of template-based pages doesn't hold up against the latest core and spam filters.

The interesting part for anyone managing data-driven sites is the traffic shift. While Google has basically ghosted the domain, ChatGPT is driving the majority of current visitors. We are moving away from traditional indexing toward a landscape of AI-driven discovery.

I'm currently re-engineering my content structure. The goal isn't just keyword density anymore; it's "Information Gain." If you want an LLM to cite your data or include it in a RAG (Retrieval-Augmented Generation) pipeline, the content has to look like it had human effort behind it rather than just being a programmatic output.

To fix my existing pages, I've been running an audit via this prompt to see where the "automated" fluff is triggering filters:

Act as an expert content strategist and SEO auditor. Analyze the following programmatic content template and the generated output. Evaluate it based on "Information Gain" and "Perceived Human Effort." 
Identify if the text follows a repetitive pattern that would trigger a spam filter or look like low-effort AI fluff. Suggest specific ways to inject unique data points or nuanced perspectives that would make this content more valuable for both Google's helpful content system and LLM retrieval-augmented generation (RAG).
Template:
[Insert Template Here]
Generated Content:
[Insert Content Here]

If you are running similar scale operations, you need to stop optimizing for bots and start optimizing for conversational intelligence. The value is in the substance, not the volume.