Yeah, I’m seeing it too, and it’s annoying as hell. We’ve had a couple client queries where the “obvious” page just gets passed over and Google/whatever pulls some random forum junk or a thin roundup instead. Not even in a “the page is bad” way — sometimes the page is fine, indexed, linked internally, all that. It just… doesn’t get chosen. Feels like the old “best page wins” thing is getting weaker and the systems are leaning hard on whatever looks easiest to map semantically. Which, honestly, is a problem because a lot of decent pages don’t phrase things the way the model wants. I’ve also noticed pages with cleaner intent matching seem to get picked even if they’re less authoritative on paper. So yeah, “good content” isn’t enough if it doesn’t line up with the query interpretation. That’s the part people keep missing when they say “just write better pages.” Not saying it’s all semantic magic, but there’s definitely some weird retrieval behavior going on. We’ve tested enough stuff now that I don’t think it’s just bad luck.
ToolDecision
Tools, compared and explained
ToolDecision
Tools, compared and explained