Has anyone actually seen consistent gains from AI-assisted internal linking at scale?

[mercer]

I’ve been testing AI-assisted internal linking on a few content sites over the last couple of months, and I’m starting to think the results depend a lot more on site structure than on the tool itself.

On paper, it sounds great: crawl the site, identify relevant anchors, insert links automatically, and improve topical coverage + crawl paths. In practice, I’ve seen three very different outcomes:

1. **On a tightly clustered niche site**
This worked reasonably well. Pages were already grouped around clear subtopics, so the AI suggestions were mostly sensible. I still had to review anchors manually, but the workflow saved time.

2. **On a broader affiliate site**
The suggestions got messy fast. The system kept forcing links where the intent was weak or the anchor text felt unnatural. It looked “optimized” from a tool perspective, but not from a user perspective.

3. **On older content with mixed quality**
This was the worst case. The AI kept linking to outdated articles that I probably should have consolidated or noindexed instead. In other words, the automation exposed structural problems rather than solving them.

My current view is that AI internal linking is useful as a **support layer**, but it’s not a replacement for manual site architecture. If your content map is weak, automation just scales the weakness.

A few things I’ve found helpful:
– Limit automation to pages in the same topical cluster
– Exclude thin or outdated posts from auto-linking
– Use exact-match anchors sparingly
– Review links on money pages manually
– Track whether the linked pages actually move in GSC, not just whether the tool says “improved coverage”

I’m curious how others are handling this.

Are you using AI for internal linking in a fully automated way, or just for suggestions?
And has anyone seen measurable ranking lifts from it, especially on affiliate sites or larger WordPress builds?

[Author]

Posts