Noindex acts on a page, nofollow acts on a link: almost all the confusion around nofollow and noindex comes from this difference in target. The noindex directive asks search engines not to show a page in their results, and Google removes it from its index even if other sites link to it. The nofollow attribute, placed on a link, tells Google that you do not wish to associate your site with the destination page, nor have Google crawl it by following the link from your site. The first controls the visibility of your own pages, the second qualifies the links you make to other sites.
Our position is fairly clear-cut. On an SME website, noindex is reserved for a handful of well-identified pages, and nofollow, with its two variants sponsored and ugc, for outbound links you do not endorse or that were paid for. The robots.txt file, which people often mix up with the other two, is never a way to hide a page. Most mistakes come from this mix-up.
Noindex, nofollow and robots.txt: three tools, three roles
Before getting to practical cases, it is worth setting out each definition as Google words it in its Search Central documentation. The words matter, because each tool works at a different stage: crawling (the robot reads the page), indexing (the page enters Google's database) or link evaluation.
Noindex: the page stays accessible, but drops out of the results
The noindex rule can be set in two ways, which have exactly the same effect. The most common is a meta tag placed in the head section of the page:
<meta name="robots" content="noindex">
This version applies to all search engines that recognise the rule. To target only Google's robots, replace robots with googlebot :
<meta name="googlebot" content="noindex">
The second way uses the HTTP header X-Robots-Tag, sent by the server along with the page. It is essential for files that are not HTML, such as a PDF, an image or a video, since you cannot slip a meta tag into them:
HTTP/1.1 200 OK
X-Robots-Tag: noindex
A visitor who knows the address can still open the page. Noindex therefore protects nothing: it simply removes the page from the results. Google's documentation also states that other search engines may interpret the rule differently, and so may keep displaying the page.
Nofollow: a hint about a link, not a ban
On a link, nofollow is written in the relattribute. Google now distinguishes three values. rel="sponsored" is used for advertising or paid links. rel="ugc" applies to links published by your users, in comments or a forum. rel="nofollow" remains the generic value, to be used when the other two do not apply.
<a href="…" rel="sponsored">Notre partenaire</a>
<a href="…" rel="ugc">Lien laissé dans un commentaire</a>
<a href="…" rel="nofollow">Site que nous ne cautionnons pas</a>
The status of these attributes changed in 2019. Previously, Google simply ignored nofollow links. On 10 September 2019, Google announced that nofollow, sponsored and ugc would from then on be treated as hints for ranking. Since 1 March 2020, nofollow has also been a hint for crawling and indexing. In practice, the current documentation says that these links are "generally" not followed, and that the linked page may still be crawled if Google discovers it another way, for example through a sitemap.
Finally, there is a page-level nofollow, in the meta robots tag. It then asks that none of the links on the page be followed. The value none is equivalent to noindex, nofollow combined.
Robots.txt: a crawling setting, not an indexing one
The robots.txt file, placed at the root of the site, tells robots which addresses they may crawl. Google presents it as a tool for managing robot traffic, mainly to avoid overloading the server. Its documentation states it plainly: it is not a way to keep a page out of Google. If other sites link to a blocked page, its address may appear in the results, without a description.
Comparison table: nofollow, noindex, robots.txt and canonical
The table below covers the four tools people confuse most often. We have added the canonical tag, because it is often the right answer where people instinctively think of noindex.
| Directive | What it does | Where it goes | When to use it | Common mistake |
|---|---|---|---|---|
| noindex | Removes the page from Google's results | Meta robots tag in the head, or header X-Robots-Tag |
Pages useful to visitors but of no interest in Google | Also blocking it in robots.txt |
| nofollow | Indicates not to associate your site with the linked page | Attribute rel of the link |
Outbound link you do not endorse | Believing the linked page will not be indexed |
| sponsored | Flags a paid or advertising link | Attribute rel of the link |
Affiliate links, sponsored article, banner | Leaving a paid link without an attribute |
| ugc | Flags a link published by a user | Attribute rel of the link |
Comments, reviews, forum | Forgetting it on open comments |
| Disallow | Prevents an address from being crawled | robots.txt file at the root of the site | Areas not worth crawling, on large sites | Using it to remove a page from Google |
| canonical | Designates the reference version of duplicate content | Link tag in the head |
Duplicates, variants of the same page | Preferring noindex to handle duplicates |

Practical cases on an SME website
Theory becomes clearer with real pages. Take a ten-person accounting firm in Lausanne, with a brochure website, a contact form, a small blog and an area where its clients upload supporting documents. This typical scenario covers most of the situations a French-speaking Swiss SME encounters.
The thank-you page after a form
After submitting the form, the visitor lands on a page saying "Thank you, we will get back to you". It has no value in Google's results, and anyone arriving there directly would have nothing to do on it. This is the textbook case for noindex. We also remove it from the sitemap, so as not to send Google two contradictory signals. The page stays crawlable: Google must be able to read it to see the rule.
Internal search results
The site's search engine generates an address for each query typed, of the form /recherche?q=tva. These pages are lists with no content of their own, and they can multiply without limit. On a site of a few hundred pages, a noindex on the results template is enough. On a much larger catalogue, a Disallow may be justified to spare crawling, but only once these pages have dropped out of the index (we come back to this below).
The client area
For the area where the accounting firm exchanges documents with its clients, neither noindex nor robots.txt is suitable. Both leave the page accessible to anyone who knows its address, and Google points out that some robots ignore robots.txt rules. Its documentation recommends password protection for files that must not appear. A serious client area therefore goes through a login, and the question of indexing no longer arises. The same reasoning applies to a staging site during a redesign: a password protects it better than a noindex, which is too often forgotten and left in place at launch, as we explain in our checklist for a redesign that keeps your SEO.
Affiliate links and sponsored content
Imagine the accounting firm recommends accounting software through an affiliate link, or publishes an article paid for by a partner. These links must carry rel="sponsored". Google states that, to avoid a possible action for link schemes, a paid link must be marked sponsored or nofollow. Both are treated the same way, but Google prefers sponsored when possible. Old links using nofollow do not need to be changed.
<a href="…" rel="sponsored">Essayer le logiciel</a>
Comments and reviews posted by your visitors
If the blog accepts comments, the links that visitors leave in them take rel="ugc". Several values can be combined, separated by a space or a comma. Google also cites rel="nofollow ugc" as a good solution for staying understood by services that do not yet recognise ugc.
<a href="…" rel="nofollow ugc">Site du commentateur</a>
Google also accepts removing the attribute for trusted contributors, to acknowledge their work. On an SME blog, where comments remain rare, we keep ugc everywhere.
PDFs and other files
An old PDF price list showing up in Google is a classic. Since the file has no headsection, the rule goes through the X-Robots-Tagheader, set on the server side. On an Apache server, Google's documentation gives this example for all the PDFs on the site:
<Files ~ "\.pdf$">
Header set X-Robots-Tag "noindex, nofollow"
</Files>
Mistakes to avoid with nofollow and noindex
These settings fit in a single line of code, which makes them easy to apply and easy to get wrong. Four mistakes come up again and again.
Combining noindex and Disallow
This is the most widespread mistake, and the most counter-intuitive. You think you are doubling the protection by adding a noindex to the page and a Disallow in robots.txt. In reality, the Disallow prevents Google from crawling the page, and therefore from reading the noindex tag. Google's documentation is explicit: if the page is blocked by robots.txt, the robot never sees the rule, and the page may still appear in the results if other pages link to it. Google also does not support a noindex rule written in the robots.txt file itself.
# À ne pas faire si la page porte un noindex
User-agent: *
Disallow: /merci/
The correct order is as follows. First add the noindex, leave the page crawlable, then wait for it to disappear from the index. Any Disallow comes only afterwards, and only if it is genuinely useful for crawling.
Believing nofollow prevents the linked page from being indexed
A nofollow removes nothing from Google. It qualifies the link, not the destination page. If that page is linked elsewhere or appears in a sitemap, Google can crawl and index it normally. To keep one of your pages out of the results, only a noindex on the page itself does the job, and this is also what Google recommends for internal links: a Disallow to avoid crawling, a noindex on a crawlable page to avoid indexing.
Putting nofollow on internal links to "sculpt" PageRank
The idea dates back to the 2000s: by placing nofollow on links to the legal notice or the contact page, people hoped to concentrate authority on commercial pages. This has not worked for a long time. In a June 2009 post, Matt Cutts, who was then working at Google, explained that the share of PageRank attributed to a nofollow link was no longer redistributed to the other links: it is simply lost. He advised against the practice and recommended letting authority flow freely within the site, with a marginal exception for links such as the shopping cart or the login page. Since 2019, as nofollow is only a hint, the calculation is even less predictable. The only real way to steer internal authority remains choosing which pages you link to, starting with the home page.
Using noindex to handle duplicates
When two addresses display the same content, the noindex reflex is tempting. Google advises against it for choosing the canonical page within a single site, since noindex removes the page from search entirely. It does not recommend robots.txt for this purpose either. A permanent redirect or a canonical tag are the two signals it considers strong, and the canonical tag remains the solution it favours.
Checking your settings in Search Console
Google's Search Console lets you check each directive without a paid tool. The page indexing report sorts non-indexed addresses by reason, and four statuses speak directly to nofollow, noindex and robots.txt.
- "Excluded by 'noindex' tag": Google found the rule and did not index the page. This is the expected result for a thank-you page, and a warning sign if the page concerned is a service page.
- "Blocked by robots.txt": crawling is forbidden, which, according to Google, does not necessarily prevent the page from being indexed by other means.
- "Indexed, though blocked by robots.txt": the page is in the index even though Google cannot read it, usually because another site links to it. If you want to remove it, Google recommends lifting the block and adding a noindex.
- "Duplicate without user-selected canonical": Google chose another page as the reference. This is a sign that a canonical tag is missing, not that a noindex is needed.
The URL Inspection tool complements the report. It shows the HTML code that Googlebot actually received, which lets you check that the noindex tag is really there, and it lets you request a new crawl. Patience is still needed: Google states that a page that is still visible has probably not been recrawled since the rule was added, and that this can take months depending on the importance of the page.
This check is one of the first verifications in our SEO support, because a forgotten noindex on a service page is enough to make it invisible. The setting itself is simple; what takes attention is knowing which pages deserve to be in Google and which do not.
If you are unsure about a particular page, or if your indexing report shows statuses you do not understand, we can go through them together in a first conversation.
Sources
- Google Search Central: blocking indexing with the noindex rule
- Google Search Central: meta robots tag and X-Robots-Tag header
- Google Search Central: qualifying outbound links
- Google Search Central: introduction to robots.txt
- Google Search Central: consolidating duplicate URLs
- Google Search Central: the evolution of nofollow and the new link attributes (2019)
- Search Console Help: page indexing report
- Matt Cutts: PageRank "sculpting" (2009)




