Comment sections went out of fashion for good reasons. Moderation is work, spam is relentless, and most of what accumulates is noise. A lot of teams turned them off and never thought about it again.
The thing worth reconsidering is that comments are one of the few places where real people ask the exact questions your content did not answer, in the exact words they would use. That is unusually valuable material for AI search, and it is sitting unused on a lot of sites that already have it.
Comments Are Question and Answer Pairs You Did Not Have to Write
The content gap analysis guide is about finding questions your content fails to address. A comment section is that data arriving unprompted.
Someone reads your guide, gets to the end, and asks the thing you left out. If you answer well, you now have a question and answer pair on the page, phrased by an actual reader rather than a marketer guessing at phrasing. Retrieval systems handle that structure well because it matches how queries arrive.
This is the same logic behind why community platforms perform strongly in AI citation, covered in the community signals guide. The difference is that on your own domain you keep the signal instead of donating it to someone else's platform.
The Liability Side Is Real
Unmoderated comments do genuine damage and it is worth being blunt about how.
Spam links pointing at whatever the spammer is selling appear on your pages under your domain. Even with nofollow applied, the visible text is now part of what a crawler reads on that page.
Volume of low quality text dilutes the page. A well written thousand word guide followed by four hundred words of "great post thanks" and "check out my site" changes the ratio of substance to noise on a page you worked hard on.
Factually wrong answers from other commenters sit on your page with your name at the top. A machine extracting from that page has no reliable way to distinguish your considered guidance from a stranger's confident mistake in the comments below it. That one bothers me more than the spam does, because spam is obvious and a plausible wrong answer is not. It also undercuts the author level credibility covered in the author entity guide, since the byline at the top implicitly owns everything underneath it.
Moderation Is the Whole Decision
Comments are worth having if you moderate them and worth turning off if you will not. There is no version where you leave them open, ignore them, and come out ahead.
What moderation means in practice: delete rather than approve anything that is not a genuine question or a substantive contribution. "Thanks, great article" is not a contribution. It is polite and it is noise, and approving it because it feels rude to delete is how comment sections turn into padding.
Answer real questions in the thread rather than only in email. An answered question is content. An unanswered question sitting under your article is a visible gap.
Correct wrong information rather than deleting it, when the wrong belief is common. A visible correction is more useful than silence, because the wrong belief exists in your readers' heads whether or not it exists on your page.
Do Not Mark Up Comments as FAQPage
The temptation is obvious and I would avoid it. FAQPage schema is meant for questions and answers the publisher stands behind, and marking up arbitrary reader comments as though they were editorial FAQ content is the kind of schema to content mismatch covered in the schema errors guide.
Comment schema exists for actual comments, using Comment type and the comment property on the parent article. It describes what the content actually is, which is the correct approach.
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Your article title",
"comment": [
{
"@type": "Comment",
"text": "Does this apply to sites using client side rendering?",
"author": { "@type": "Person", "name": "Reader name" },
"dateCreated": "2026-07-14"
}
]
}If a comment question is good enough that you want it in your FAQ, promote it. Rewrite it into the article body or the FAQ section as editorial content you are standing behind, and mark that up properly. That is a genuine content improvement rather than a markup trick.
Pagination Applies Here Too
Long comment threads often load in batches on scroll, which puts them in exactly the situation described in the pagination guide. If comments load via JavaScript after page render, a non rendering crawler sees none of them.
That is either a problem or a feature depending on whether your comments are worth reading. If you moderate well and the threads carry real substance, get them into the server rendered HTML. If you do not moderate, client side loading is accidentally protecting you.
Third party comment platforms almost universally load client side, which means their content is usually invisible to crawlers regardless of quality. Worth knowing before you attribute value to it.
My Actual Recommendation
For most B2B and SaaS sites, close comments and put the effort into structured FAQ sections instead. The moderation cost is real, the volume of genuine questions on a typical B2B blog is low, and a curated FAQ you control is a better asset than a thread you have to police.
For sites with genuine community, technical documentation, or anything where readers routinely hit edge cases you did not anticipate, keep them and moderate them properly. The question volume justifies the work and the edge case answers are content you would never have thought to write.
What I would not do is keep them open out of inertia while nobody watches them.
Checking What Is Actually There
Load one of your older popular articles with JavaScript disabled and see whether comments appear. That tells you whether any of this is reaching crawlers at all.
The NotionCue AI Crawler Audit reports what crawlers receive rather than what renders, which settles the question for pages where comments are a meaningful share of the text.
Start your free NotionCue trial and check an article with a long comment thread. The answer determines whether moderation effort is worth anything from an AEO standpoint.
If you are turning comments off, redirect or remove the standalone comment permalink URLs some platforms generate. Those are thin pages that outlive the feature and nobody remembers they exist.
Common Questions
Does nofollow on comment links solve the spam problem?
It handles link equity and does nothing about the visible text sitting on your page. Removal is the fix, not attributes.
Are comments worth keeping purely for freshness signals?
No. A page whose only recent change is new comments has not been genuinely updated, and treating comment activity as freshness is the kind of hollow signal the content decay guide warns against.
What about comments on documentation specifically?
Usually the strongest case for keeping them. Documentation readers hit real edge cases, and the resulting threads are frequently the only place a specific problem is discussed anywhere.