Is AI Content Actually Helping Your Rankings, or Just Making Production Easier?

[mercer]

I’ve been testing a few AI-assisted content workflows across smaller niche sites over the last several months, and I keep coming back to the same question: are we actually improving rankings, or are we just making it cheaper and faster to publish?

My current setup is a mix of AI drafting, manual fact-checking, internal linking, and some light on-page optimization. In theory, it should be a solid system. In practice, the results are mixed.

What I’m seeing so far:
– AI helps a lot with scale and consistency
– It’s useful for outlines, FAQs, comparison tables, and first-draft product sections
– It still struggles with nuance, especially for YMYL-adjacent topics and anything that needs real firsthand insight
– Pages that get a human pass for intent matching and internal linking tend to perform better than “publish and pray” content

One thing I’m paying more attention to now is whether the content is actually satisfying the query or just looking complete. Those are not the same thing. A lot of AI content can appear polished while still missing the practical angle that users and Google seem to reward.

I’m also seeing internal linking matter more than people give it credit for. On a few sites, pages with strong supporting links from related articles started moving faster than pages that were arguably better written but isolated. That makes me think content quality is only part of the equation. Topical structure still matters a lot.

Curious how others are handling this:
– Are you using AI mainly for speed, or are you seeing direct ranking gains?
– Do you edit heavily, or rely on lighter human review?
– Have you noticed better results from AI content when paired with stronger internal linking?
– For affiliate sites, are you seeing better conversion from AI-assisted pages or from more manual, experience-based content?

My take is that AI is a good production layer, but not a replacement for site strategy. If the topical map is weak, the internal links are random, or the page doesn’t add anything beyond what’s already ranking, the content usually underperforms no matter how clean it reads.

Would be interested to hear what’s actually working in real projects, not just theory.

[Author]

Posts