Fix noarchive and nocache: Bing Copilot Controls
Your page carries a noarchive or nocache robots directive. Since September 2023 Microsoft has used these two tokens to control how Bing Chat, the product now called Copilot, may use a page: noarchive keeps the page out of Copilot answers entirely, and nocache lets Copilot use only the URL, title and snippet rather than the page's own content. Neither affects Google, which stopped using noarchive when it retired cached links, and neither removes the page from any index. If the directive was left behind by an old caching or archiving setting, remove it; if you meant it, nothing here needs fixing.
What this means
This is a notice, not an error. The audit found one of two robots directives on the page:
- noarchive — Microsoft says content tagged NOARCHIVE "will not be included in Bing Chat answers", and will not be linked to in them either (Bing Webmaster Blog, 2023-09-22, read 2026-09-23).
- nocache — content with this tag may still appear in a Copilot answer, but Microsoft says it will "only display URL/Snippet/Title in the answer" (same source). The page's own wording is not used.
Both can be set in three places, and the audit reports which one it found: a <meta name="robots"> tag, a <meta name="bingbot"> tag aimed at Bing specifically, or an X-Robots-Tag HTTP response header set by your server or CDN. A directive in the header is invisible in the page source, which is the usual reason people cannot find where it comes from.
Both tokens predate AI answers. noarchive originally meant "do not show a cached copy of this page" and was widely pasted into templates for that reason. Microsoft gave it a second, much larger meaning in September 2023; Google went the other way and retired it. Google's robots meta tag documentation now says noarchive is no longer used by Google Search, because the cached-link feature no longer exists (updated 2026-03-24, read 2026-09-23). So a noarchive tag you added years ago to hide a cached copy is, today, doing nothing in Google and quietly excluding you from Copilot.
Why it matters
Copilot is the answer surface built on the Bing index, and Microsoft is the one major operator that publishes explicit, named controls for it. That cuts both ways: it is the only AI answer engine you can deliberately opt out of with a meta tag, and the only one you can opt out of by accident.
If you want to be cited in Copilot, noarchive is the single directive that prevents it outright. Microsoft's wording is that tagged content is not included in the answers and not linked to in them, so you lose both the mention and the click. nocache is softer but still costly: the answer can name your page and show its title and snippet, but Copilot does not draw on the page's own content, so the substance of the answer comes from somewhere else.
Be clear about the scope, because these directives are often described as an "AI opt-out" and they are not. They are Microsoft controls. They do not govern OpenAI's GPTBot or OAI-SearchBot, Anthropic's ClaudeBot, PerplexityBot, or Google's AI Overviews and AI Mode — those are controlled by robots.txt user-agent rules and, for Google, by the nosnippet and max-snippet directives. Removing noarchive changes nothing in Google, and adding it blocks nothing outside Microsoft.
Neither directive is an indexing or ranking control. The page stays in the index and stays eligible to rank; noindex is the directive that removes a page from search. That is why this is a notice: it is a deliberate-or-not choice about one engine's answers, not a defect.
How to fix it
- 1
Decide whether you meant it
Do this first, because the correct action may be none. Publishers who license their content, sites under an editorial policy that keeps them out of AI answers, and pages you simply do not want quoted are legitimate places for noarchive. If that is a deliberate decision, leave it and treat this notice as confirmation that the directive is working. The rest of these steps are for the far more common case: a token inherited from a template, an SEO plugin default, or an old instruction to suppress a cached copy that no longer exists anywhere.
- 2
Find where the directive is set
The audit's finding names the source it found, and there are three. In the HTML, view source and search for
noarchiveandnocache; they can sit in a<meta name="robots">tag that applies to every crawler, or in a<meta name="bingbot">tag that applies only to Bing and will not be found by anyone searching forname="robots". In the response headers, runcurl -sI https://example.com/page | grep -i x-robots-tagor open the Network tab and read the response headers for the document request. Check all three: a header applies on top of the meta tags, so removing the tag alone can leave the directive in force. - 3
Remove the token from the meta tag
Delete just the offending token, not the whole tag.
<meta name="robots" content="index, follow, noarchive">becomes<meta name="robots" content="index, follow">, or you can drop the tag entirely, since the defaults already allow indexing and following. If the token sits in a<meta name="bingbot">tag and nothing else is in it, remove the whole tag. Do not replace noarchive with noindex or nosnippet: those are different directives with much larger effects. - 4
Remove it from the X-Robots-Tag header
If the directive is in the header, editing the HTML will not help. In Apache, look for
Header set X-Robots-Tagin.htaccessor the vhost. In Nginx, look foradd_header X-Robots-Tag. On Cloudflare, check Rules, then Transform Rules, then Modify Response Header, and any Workers that set response headers. Remove the token, or scope the rule so it applies only to the paths you genuinely want excluded, then purge the cache so corrected headers are served. - 5
Confirm the change and give Bing time to recrawl
Fetch the page again and confirm noarchive and nocache are gone from both the HTML and the response headers, then re-run this audit to clear the notice. Bing will not act on the change until it recrawls: use the URL Inspection and Submit URL tools in Bing Webmaster Tools to ask for a fresh crawl, and check the Crawl Information there to confirm what Bing currently holds. There is no way to verify Copilot eligibility directly, so the crawl is the observable part; whether any given answer cites you is not something you or this audit can force.
Example
<!-- FLAGGED: keeps the page out of Microsoft Copilot answers entirely -->
<meta name="robots" content="index, follow, noarchive">
<!-- FLAGGED: aimed at Bing only - a search for name="robots" will miss it -->
<meta name="bingbot" content="noarchive">
<!-- FLAGGED: Copilot may cite the page, but shows only URL / title / snippet -->
<meta name="robots" content="index, follow, nocache">
<!-- FLAGGED as an HTTP header, invisible in the page source: -->
<!-- X-Robots-Tag: noarchive -->
<!-- FIX: drop the token, keep the rest -->
<meta name="robots" content="index, follow">
<!-- or omit the tag entirely; the defaults already allow indexing and linking -->The three places noarchive and nocache hide, and the permissive equivalent. Remove only the offending token - noindex and nosnippet are different directives with far larger effects.
Platform-specific steps
Both plugins expose the robots meta tokens per post and as site-wide defaults. In Yoast, open the post's SEO panel, go to Advanced, and review the meta robots settings; the advanced robots fields are where a stray noarchive usually lives. In Rank Math, edit the post, open the Advanced tab of the SEO panel and check the Robots Meta boxes, then check the same defaults under Titles & Meta for the post type. If the token appears across a whole content type rather than one post, it is the site-wide default, not the post.
Grep your shared layout, head component and any SEO helper for noarchive and nocache. In the Next.js App Router the robots metadata export is the usual source: robots: { index: true, follow: true } emits no archive token, while a hand-written <meta name="robots"> in a layout will override your expectations. Also grep middleware and any API or server code that sets response headers, since an X-Robots-Tag there applies on top of whatever the page renders.
Edge rules are the most common hiding place for a header-set directive, because nothing in your repository mentions it. On Cloudflare, check Rules, then Transform Rules, then Modify Response Header, and review any Workers that modify the response. On other CDNs look for the equivalent response-header rule set. Remove the token or scope the rule to the paths you actually want excluded, then purge the cache so corrected headers are served.
Frequently asked
No. noarchive is not an indexing or ranking directive. Your page stays in the index and stays eligible to rank. What Microsoft says it changes is Copilot: tagged content is not included in Bing Chat answers and is not linked to in them (Bing Webmaster Blog, 2023-09-22, read 2026-09-23). In Google it does nothing at all any more, because Google retired cached links and its documentation now says noarchive is no longer used by Google Search (robots meta tag documentation, updated 2026-03-24). If you want a page out of search entirely, the directive is noindex.
noarchive is the harder block: Microsoft says the content will not be included in Bing Chat answers and will not be linked to in them. nocache is partial: the page may appear in an answer, but Microsoft says only the URL, title and snippet are displayed, so Copilot does not draw on the page's own content. In short, noarchive removes you from the answer, nocache reduces you to a citation line. Both come from the same Microsoft announcement of 2023-09-22 (read 2026-09-23).
No, and neither does adding it. These are Microsoft directives for Bing and Copilot. OpenAI, Anthropic and Perplexity are controlled through robots.txt rules for their own crawler user-agents, such as GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot. Google's AI Overviews and AI Mode are governed by Googlebot access plus the nosnippet and max-snippet directives, which are a separate check in this audit. Treat noarchive and nocache as a Microsoft setting, not as a general AI opt-out.
Two likely places. First, an X-Robots-Tag HTTP response header, which never appears in the HTML: check it with curl -sI or your browser's Network tab, and look at your server config, your CDN's response-header rules and any edge Worker. Second, a <meta name="bingbot"> tag rather than <meta name="robots">, which a search for name="robots" will miss. This audit names the exact source it found in the finding itself, so start there.
Only if your target is Microsoft Copilot specifically and you accept losing the citation and the link with it. It is a real, documented control and it is honoured, but its reach stops at Microsoft; every other engine needs its own robots.txt rules or directives, and no single tag covers them all. It is also a blunt instrument: there is no setting that keeps the link while dropping the content. Most sites that want AI visibility should remove it; sites with a licensing or editorial reason to stay out should keep it and ignore this notice.
Does your site have this issue?
Run a free, AI-powered audit and we’ll flag this and 150+ other checks in about a minute. No signup.