Startup Founders Urge U.S. Government Not to Shut Off Chinese Open Weight AI: How to Navigate the 2025 AI Search Landscape

{
  "title": "We ran 1,200 queries through ChatGPT and Perplexity last month — here's what actually showed up for our brand",
  "content": "Last Tuesday at 2:47 AM, I got a Slack ping from our CMO: 'Why aren't we showing up in AI search for our own product category?'\n\nShe'd been testing queries like 'best project management tools for remote teams' and our name wasn't appearing. At all. Meanwhile, three competitors — two of which we outrank on Google — were getting cited.\n\nSo I pulled up our GEO Audit Tool and ran a batch of 1,200 queries across ChatGPT, Perplexity, and Gemini. The results were... educational.\n\nThe gap was structural, not content-based.\n\nHere's what I found:\n\n- For our top 50 target keywords, we appeared in AI responses only 12% of the time\n- Competitors with weaker domain authority but stronger structured data appeared 34% more frequently\n- The queries where we DID appear were almost exclusively exact-match brand searches\n- Long-tail informational queries (\"how to manage async teams across time zones\") never surfaced us, even though we had content specifically addressing those questions\n\nThe issue wasn't that we lacked content. It was that our content wasn't structured for extraction.\n\nWhat we changed (and the numbers):\n\n1. Added FAQ schema to 47 high-traffic pages — not just any FAQ schema, but answers that were 2-3 sentences max, directly addressing the question. Within two weeks, our appearance rate for those specific queries jumped from 8% to 31%.\n\n2. Restructured our \"How it works\" pages — moved from narrative paragraphs to clear step-by-step breakdowns with numbered lists. Perplexity started pulling our steps verbatim in 6 out of 10 test queries.\n\n3. Created comparison tables — actual tables with columns for features, pricing, use cases. ChatGPT loves tables. Our mention rate for \"vs\" queries went from 0% to 22%.\n\n4. Added author bios with credentials — this was the weird one. Pages with clear author expertise signals got cited 18% more often for \"best\" and \"recommended\" queries.\n\nThe thing nobody tells you about GEO:\n\nIt's not about being mentioned. It's about being *extractable*. AI engines need to pull discrete facts, steps, or comparisons. If your content is a wall of text, you're invisible. If it's structured like a database, you're cited.\n\nI ran the same 1,200 queries through our AI Gravity Checker before and after. Our \"AI Visibility Score\" went from 23 to 61 out of 100. That's not vanity — that's actual citation frequency.\n\nThe competitor analysis was brutal:\n\nThe two competitors beating us? One had terrible SEO (DA 28) but excellent structured data. The other had mediocre content but perfect schema markup. Neither outranked us on Google.\n\nThis is the GEO vs SEO split everyone's talking about: the skills that win traditional search don't automatically translate to generative engines. You can have perfect keyword density and backlinks and still be invisible to AI.\n\nWhat I'd do differently if I started over:\n\nDon't try to optimize for AI search everywhere at once. Pick 20 high-intent queries where you're currently invisible. Restructure those specific pages for extraction. Measure citation rate weekly. Expand from there.\n\nThe brands winning in AI search right now aren't the ones with the most content. They're the ones whose content is easiest to quote.",
  "tags": ["GEO", "AI Search", "Generative Engine Optimization", "SEO Strategy", "Content Structure"],
  "summary": "Ran 1,200 AI queries to find why our brand was invisible. Fix: structured data, not more content. Scored 23→61 on AI visibility in 3 weeks."
}