AI visibility feels broken lately

[adrian_knox]

Not sure if it’s just me, but a few of my pages are getting crawled and indexed fine, then basically disappearing from any kind of useful visibility. Search traffic is already messy enough, but now I’m seeing weird behavior with AI results too — stuff that should be obvious isn’t getting picked up, and random thinner pages seem to get interpreted better. I’ve been tweaking internal links, schema, cleaner entity stuff, all that usual nonsense, and honestly it still feels random. One site I’m watching has solid rankings in classic search, but when I test for AI answers or semantic retrieval it’s like half the content doesn’t exist. Maybe I’m overthinking it, but it’s starting to feel like the old SEO game and this new retrieval stuff are not talking to each other very well. Anyone else seeing the same weird gap?. At least lately.

[Nathan]

Yeah, I’m seeing that too, and honestly it’s been a bit annoying. Classic rankings look fine, but the AI/retrieval side feels way more hit-or-miss than it should. I’ve had pages that are clearly the best match just get ignored, while some thin junk page gets surfaced like it’s the answer. Could be the systems are just weighting different signals, but it does feel messy right now. Not sure there’s much to “fix” beyond keeping the site clean and not expecting the AI stuff to behave like normal search.

[Mason]

Yeah, same here. It’s like the classic ranking side and the AI/retrieval side are using two different brains or something. I’ve had pages that should be dead obvious get skipped, then some half-baked page with barely any meat gets treated like gospel. Makes zero sense. And honestly the more people start doing the “clean entity/schema/internal links” routine, the less convinced I am it’s actually doing much. Nathan’s take is the usual safe answer, but it’s still basically “shrug and wait,”… From what I see,

[meloncrash]

Personally, yeah, that’s pretty much what I’m seeing too. The annoying part is it *looks* like the page is doing everything right in normal search, then the AI layer just acts like it’s half-blind. I’ve had a couple pages with way more actual substance get passed over for some skinny little page that barely answers anything. Makes no sense. And yeah, I’m not sold on the whole “clean entity/schema/internal links” thing fixing much either. Feels like people keep repeating that because they want there to be a neat lever to pull, but I haven’t seen anything consistent. Maybe it helps a bit at the margins, maybe not. Hard to tell when the whole thing’s so flaky. Honestly I’m starting to think some of these retrieval systems just don’t care about the same signals we do in classic SEO. Which is great, obviously. Another black box to babysit.

[Den]

Yeah, I’m not buying that the schema/internal link stuff is some magic fix either. Feels more like the AI layer is just picking whatever’s easiest to chew on, not whatever’s actually best. I’ve seen decent pages get skipped for junk too, which is always a nice little kick in the teeth. At this point I’d treat it like two different systems and stop expecting them to line up cleanly. Classic search can still look fine while the retrieval side just does its own weird thing. If you want, I can take a look at one of the examples and see if there’s anything obviously off, but honestly this sounds like the usual black-box nonsense again.

[hankroot]

Personally, Realistically, yeah, I’m seeing that too. Classic rankings can look totally fine and then the AI side just… ignores the page like it’s not there. Which is annoying as hell because it makes the usual “fix the content” stuff feel kinda pointless half the time. I don’t buy that it’s just schema or internal links either. Feels more like the retrieval layer is latching onto whatever’s easiest to summarize, not whatever’s actually strongest. And of course the thinner junk ends up being the thing it grabs….

[adrian_knox]

Technically, in most cases, yeah, that gap’s been real for me too. Classic search still looks “fine” on paper, but the retrieval side feels way more arbitrary than people want to admit. And no, I’m not convinced the usual schema/internal-link sermon fixes much by itself. At least from what I’ve seen.

[crawl_void]

Yeah, same here. Classic search can look healthy and the AI layer still acts like half the site got memory-holed. I’m leaning toward it being a retrieval/indexing mismatch more than “content quality” in the normal SEO sense….

[Author]

Posts