Nofollow on Internal Links: Almost Always a Mistake

Putting rel="nofollow" on internal links is almost always a mistake: it doesn't conserve link equity for your other links, it can suppress crawling and indexing of your own pages, and the "PageRank sculpting" strategy it descends from stopped working in 2009. The equity a nofollowed link would have carried simply evaporates; it isn't redistributed. There are a few narrow cases where nofollowing an internal link is defensible, and even those usually have better tools.

What nofollow actually does today

rel="nofollow" began in 2005 as a way to mark links a site didn't editorially vouch for: primarily comment spam. In September 2019, Google announced a significant reframing: nofollow (along with the new rel="sponsored" and rel="ugc") became a hint rather than a directive. Google may choose to ignore it for ranking purposes, and since March 2020 may ignore it for crawling and indexing too.

So the modern behavior on an internal link is the worst of both worlds:

• For equity: Google generally does not pass PageRank through nofollowed links. The value doesn't route around them; it's dropped.
• For crawling: Googlebot generally won't follow the link, so if it's the page's only path, the page may go undiscovered or stale.
• For certainty: because it's a hint, you can't even rely on the blocking behavior. You get unpredictability and signal loss.

You've told Google "I don't vouch for this page", about your own page.

Why PageRank sculpting died

To understand why nofollowing internal links feels intuitive to people, you need the history.

Under the original PageRank model, a page divides its outgoing equity among its links. Around 2005–2009, SEOs realized: if a page has 10 links and you nofollow 5, the remaining 5 each get a doubled share. So sites nofollowed links to privacy policies, login pages, and terms pages to "sculpt" flow toward money pages. It worked, Google confirmed it worked, and it became standard advice.

Then in 2009, Google changed the calculation, and Matt Cutts described the new model publicly: nofollowed links still count in the divisor. A page with 10 links and 5 nofollows now splits equity 10 ways and discards the 5 shares assigned to nofollowed links. Sculpting stopped concentrating equity and started incinerating it.

That was seventeen years ago, and the change has never been reversed. Yet sculpting advice still circulates, plugins still ship "nofollow this link" checkboxes, and audits still turn up sites nofollowing their own category pages in 2026. The intuition ("hoard the juice") outlived the mechanics by a generation.

"But I don't want that page ranking"

The usual motivation for internal nofollow is a page you don't want in search: cart, login, thank-you pages, filtered category permutations. Legitimate goal, wrong tool. Match the tool to the actual problem:

Note the mechanics: noindex lets the page be crawled but excluded from results, while links to it still resolve normally. robots.txt stops the crawl before it starts, which is what you want for genuinely infinite spaces. Nofollow does neither job reliably (remember, it's a hint) while still burning the equity.

There's also a hybrid trap worth knowing: a long-term noindex page eventually has its links treated as nofollow, so don't count on noindexed hub pages to pass equity onward either.

The rare defensible cases

Honesty requires listing where internal nofollow is at least arguable:

1. Login/account URLs with session parameters, though robots.txt handles this cleaner.
2. Faceted navigation generating millions of filter combinations, where you're layering defenses (robots.txt patterns plus nofollow on filter links) during a crawl-budget crisis on a very large site.
3. Internal search result URLs linked from within content; again, robots.txt is the standard fix.

Even in these cases, nofollow is the backup tool, not the primary one. If you find yourself reaching for internal nofollow anywhere else, category pages, tag pages, author archives, even your pricing page (yes, this happens), stop and identify which real problem you're solving, because nofollow solves none of them well. The contrast with outbound external links is total: there, nofollow/sponsored/ugc remain the correct, Google-recommended markup for links you don't editorially endorse. The two sides of that divide are mapped in internal vs external links.

How accidental nofollows happen

Most internal nofollows in the wild weren't decisions; they're artifacts:

• Sitewide meta tags. <meta name="robots" content="nofollow"> in a template nofollows every link on every page using it. This one line has cratered sites. It sometimes ships from staging, where it was added alongside noindex to keep the staging site out of Google, then survives the launch checklist.
• Plugin defaults. SEO and affiliate-link plugins with "nofollow external links" options that misclassify internal links (protocol-relative URLs, www vs non-www mismatches, links to subdomains).
• UGC handling applied too broadly. A forum or comment system that nofollows all links in user content (correct), but the same sanitizer also processes staff-written content.
• Copy-paste inheritance. An editor copies a link (with its rel attribute) from a sponsored post into an organic one, and the attribute propagates through years of "duplicate this post" workflows.
• Domain migrations. Links marked external-and-nofollowed under the old domain become internal after a rebrand consolidates sites, but keep their attributes.

Auditing for internal nofollows

The audit is mechanical and worth doing once per quarter on large sites, once ever on small ones:

1. Crawl the site with any crawler that records link attributes (Screaming Frog: Internal report, filter rel="nofollow"; most crawlers expose this).
2. Check templates for meta robots. Search your codebase or view-source on each distinct template for content="nofollow" and X-Robots-Tag headers; a template-level nofollow won't show up as a per-link attribute.
3. Classify every hit. For each nofollowed internal link: was this deliberate? Does it match one of the three defensible cases? If not, it's a bug.
4. Fix at the source. Remove the attribute in the template/plugin/content, and fix the generator, not just the instances, or they'll be back next quarter.
5. Re-crawl and diff. Confirm the nofollow count dropped to your small deliberate set (ideally: zero).

While you're in the crawl data, the same pass usually surfaces the related pathologies: orphaned pages, equity pooling in dead ends, one-phrase anchor profiles. A full internal link audit covers nofollow detection alongside the rest of the graph analysis in one crawl, which is less tedious than assembling the picture from four separate exports.

The better mental model

The impulse behind internal nofollow is real and worth honoring: not all of my pages deserve equal internal emphasis. Correct! But the lever for that is link placement and structure, not link attributes. You emphasize pages by linking to them more, from stronger and shallower pages, with better anchors, and de-emphasize pages by linking to them less, not by linking to them with poisoned links.

That's the whole discipline of internal linking: deciding which pages your site endorses, and encoding those decisions as contextual links. Equity routing through structure is covered in depth in how link equity actually flows, and the end-to-end system (auditing, structuring, anchoring, maintaining) is in the complete internal linking guide.

Every internal link should be a link you mean. If you mean it, follow it. If you don't mean it, the fix is editorial, not an attribute.