meta name="keywords"

In the mid-1990s a search engine had very little to go on, so it asked the page to describe itself. <meta name="keywords"> was that channel: a comma-separated list of the terms a document wanted to be found under. It worked well enough to be worth lying about, and within a few years the field was so comprehensively stuffed that engines stopped reading it. The tag stayed in the boilerplate anyway.

Why avoid

The unusual thing about this rule is that the specification which defines the field also documents why it is worthless. keywords is a current standard metadata name in the HTML Standard — not obsolete, not removed — and the entry defining it carries this note:

Many search engines do not consider such keywords, because this feature has historically been used unreliably and even misleadingly as a way to spam search engine results in a way that is not helpful for users.

Both engines that matter say the same in their own words. Google published Google does not use the keywords meta tag in web ranking in September 2009, stating that its web search disregards the tag entirely; that has never been walked back. Bing’s senior product manager for webmaster outreach wrote in 2014 that the tag “flat lined years ago as a booster” and that it is “dead in terms of SEO value”.

So the element is inert, but it is not free. It is invisible to your readers and plainly visible in view-source to everyone else, which means the list of terms you hope to rank for is published to your competitors at no cost to them. And because nothing reads it, nothing tells you when it goes stale: a hand-maintained list that no test covers and no crawler consumes drifts away from what the page actually says immediately and silently.

Use instead

Delete the element. Nothing replaces it, because nothing was being done.

To describe a page to a search engine or a social card, use the two things that are actually read:

<title>How to deadhead roses</title>
<meta name="description" content="A short, accurate, human-written summary of the page.">

For machine-readable topical metadata, use structured data, which search engines document and consume:

<script type="application/ld+json">
{ "@context": "https://schema.org", "@type": "Article", "about": "Gardening" }
</script>

Detectability

Fully detectable. The match is a single attribute value with no context that changes the verdict.

The selector uses = rather than ~=, which is the opposite of link/shortcut-icon and link/image-src and is deliberate: rel is a space-separated set of tokens, so a substring of it has to be matched as a token, but the name attribute of <meta> is a single value. keywords there is the whole attribute or it is a different metadata name. The i flag covers Keywords and KEYWORDS, which the HTML Standard treats as the same name.

The fix removes the element. No browser acts on it, so deletion cannot change how the page renders or behaves. One caveat is on the record rather than hidden: Bing’s 2014 post allows that the tag may retain some use for contextual ad systems, and a handful of site-internal search indexes read it. Neither is default behaviour anywhere — both are bespoke per-site integrations — and --fix is opt-in, so the op stands.

Resources

• HTML Standard — standard metadata names — defines the field, and carries the note quoted above about search engines not considering it.
• Google Search Central — “Google does not use the keywords meta tag in web ranking” (September 2009) — the primary vendor statement, still published under Google’s current documentation domain.
• Bing Webmaster Blog — Duane Forrester, “Blame The Meta Keyword Tag” (4 October 2014) — the second engine, in its own words, including the contextual-ad caveat noted above.
• Search Engine Land — “The Meta Keywords Tag Lives At Bing & Why Only Spammers Should Use It” (2011) — documents the exchange in which Bing clarified the tag as a spam-detection signal rather than a ranking one.