Does Google Crawl Nofollow Links?

Does Google crawl rel="nofollow" links? It is one of the oldest questions in technical SEO, and the answer has changed materially since it was first put to the test. Back in 2015, the Search Laboratory team ran a controlled crawl experiment that produced a clean, confident result. Almost a decade on, Google’s own treatment of the attribute has shifted, so the practical advice has moved with it. This article revisits the original experiment as a piece of SEO history, then brings you fully up to date with how nofollow actually behaves today.

The original 2015 experiment

To test whether Googlebot would follow a nofollowed link, the team built two clean test domains, each with a deliberately simple structure. Every site had a homepage, a /page-c/ reachable only through a normal followed link, and a /page-b/ reachable only through a rel="nofollow" link. Because the test pages were not linked from anywhere else and had no inbound links, the only route Google had to discover them was the link on the homepage.

Method

Both sites were initially blocked from all search engines in robots.txt while the pages and links were created.
The sites were then unblocked and submitted through Google Search Console to prompt a crawl.
Googlebot activity was tracked through raw server access logs across roughly a month, beginning 20 July 2015.
The homepage was later resubmitted to nudge further crawling and confirm the pattern held.

The results

The log data was striking. On the first site, the homepage was crawled 69 times and the followed /page-c/ was crawled 8 times — but the nofollowed /page-b/ was crawled zero times. The second site mirrored it exactly: 39 homepage crawls, 4 crawls of the followed page, and again zero crawls of the nofollowed page. From this, the conclusion in 2015 was unambiguous: Google did not crawl nofollow links, and the attribute could therefore be used to shape crawl budget and conserve indexation.

It was a tidy, well-controlled study, and at the time it reflected how the attribute was widely understood. This kind of log-led, hypothesis-driven testing is exactly the approach our data science and analytics work is built on — and it is why the finding is worth preserving rather than quietly deleting.

What changed: nofollow became a hint in 2020

Here is the part that matters for anyone reading this in the present day. On 1 March 2020, Google changed how it treats the nofollow attribute for crawling and indexing. Where nofollow had been a strict directive — a hard instruction Google would not cross — it is now a hint. Google may still choose to crawl a nofollowed link, and may use it for ranking and indexing purposes, if its systems judge that doing so is worthwhile.

In other words, the clean 2015 result no longer describes current behaviour. A nofollowed link is not a guaranteed wall any more. You should assume Google can see, crawl and potentially index the destination, and plan accordingly.

The newer link attributes: sponsored and ugc

Alongside the 2020 change, Google introduced two more specific values so you can describe why a link is qualified, rather than lumping everything under one attribute:

rel="sponsored" — for paid or advertising links, including affiliate links and any placement you have paid for.
rel="ugc" — for user-generated content, such as links inside comments and forum posts.
rel="nofollow" — still valid as a general-purpose value when you simply do not want to vouch for a link and none of the more specific values fit.

You can combine values where it is accurate to do so — for example, a paid link inside user-generated content could carry rel="sponsored ugc". All three are treated as hints for crawling and indexing.

What this means for your SEO today

The practical takeaways are very different from 2015:

Do not rely on nofollow to control crawl budget. If you genuinely need to keep Google away from a URL, use robots.txt to block crawling, or use the URL Inspection and parameter-handling tools — not the nofollow attribute.
Do not rely on nofollow to keep a page out of the index. To prevent indexing, use a noindex meta robots tag (and make sure the page is crawlable so Google can actually see it).
Use the right attribute for the job. Tag paid and affiliate links as sponsored, and tag comment and forum links as ugc. This is the cleanest way to stay on the right side of Google’s link spam guidelines.
Treat all three as honesty signals, not levers. Their job is to describe the nature of a link accurately — they are not a reliable mechanism for steering the crawler.

The broader lesson is one Search Laboratory always championed and that lives on at Havas Market: test your assumptions, but keep re-testing them. An SEO truth from 2015 can quietly become a 2020 myth, and the agencies that win are the ones that notice. You can read more about Search Laboratory’s heritage and where its specialisms sit now over on our Search Laboratory hub.

Want a second opinion on your link strategy?

If you are unsure whether your nofollow, sponsored and ugc attributes are set up correctly — or whether your crawl budget is being spent where it should be — our team can audit it and show you the evidence. The same data-led testing mindset behind this experiment now powers performance and commerce media work across the Search Laboratory hub at Havas Market.

Speak to our team