I try to hold my opinions about this category loosely, because it's moving fast enough that being wrong monthly is basically guaranteed. Here are three specific things I believed at the start of this quarter that turned out to be wrong, or at least incomplete, once I started running real audits.
1. I thought reputation was the bottleneck. It's structure.
Going in, I assumed the businesses struggling with AI visibility mostly had a reputation problem — not enough reviews, not enough press, not enough credibility. What I actually found, case after case, was the opposite: genuinely excellent businesses with real credentials that were simply unreadable to a crawler. Arros QD had seven Michelin stars and was still being described by TripAdvisor instead of itself. The reputation was never the problem.
2. I thought this was mostly a content problem. It's often a single technical setting.
I expected most of the work to be rewriting copy for extractability. Sometimes it is. But the single biggest fix I've seen this quarter wasn't content at all — it was a server response header. White Dream Villas had spent nearly a year with an SEO agency writing content that Google and every AI crawler were never even allowed to read, because of one line in the server config. I underestimated how often the real blocker is invisible and mechanical, not creative.
3. I thought volume of content mattered more than it does.
I came into this expecting "more pages" to be a meaningful lever on its own. It isn't. What actually moves an AI engine's citation behavior is depth and structure on the pages that matter, not page count. A business with ten thin pages does worse than a business with three that are genuinely comprehensive and correctly marked up.
Where that leaves me now
I'm more convinced than I was three months ago that this gap closes fast for the right businesses, and more humble about assuming I know which lever matters most before I've actually run the audit. Chad exists because I kept needing to re-run this exact diagnostic by hand.