WordPress Meta Keywords: Do They Still Matter in 2026?

You are in the WordPress editor, looking at a field labeled keywords, and you want to know whether leaving it blank costs you anything.

It doesn’t. WordPress meta keywords carry no positive ranking value on any search engine worth your traffic. Google stopped using the keywords meta tag in September 2009 and has never reversed it. Yandex is the only mainstream engine that still says it reads the tag. Bing reads it too, just not the way you’d want: Bing has said for over a decade that it treats the tag as a way to spot spammers, not a way to rank pages.

That last part is the reason this field deserves five minutes instead of zero.

Most people handling meta keywords in WordPress land in one of two camps. The first fills the field on every post, twelve terms at a time, because a plugin showed them a box and a box feels like work. The second decided years ago that the tag was dead, stopped thinking about it, and never checked whether their site is still shipping a stuffed keywords tag inherited from a plugin someone configured in 2013.

The first group wastes time. The second group is carrying a risk it can’t see.

The tag isn’t a problem because you use it. It’s a problem because you stopped looking at it.

One line of HTML, unchanged since the 1990s

The tag lives in the head of an HTML document as a comma-separated list of terms the page is supposedly about:

<meta name="keywords" content="wordpress, meta keywords, seo plugin">

Nothing about that markup has changed since the 1990s, which is the whole story in one sentence.

The original design made sense for the search engines of that era. AltaVista, Excite, Lycos, and Infoseek had no reliable way to judge what a page was about beyond the words on it, so they let publishers declare it. Publishers declared whatever paid. Within a few years the tag was carrying competitor brand names, unrelated high-volume terms, and misspellings of celebrity names on pages selling printer cartridges.

A signal that anybody can set to anything, at zero cost, with no way to verify it, is not a signal. It’s a wish list.

Google ignores the tag. Bing watches it.

These two facts get collapsed into “no search engine uses meta keywords,” and the collapse is where people get into trouble.

Google’s position is the simple one. On 21 September 2009, Google published Google does not use the keywords meta tag in web ranking on its official Search Central blog. The answer to whether the tag ever influenced Google web search was, in a word, no. Google has restated it steadily since, and no version of the ranking system has brought it back. Stuffing a keywords tag on a WordPress post buys you exactly nothing in Google.

Bing’s position is more interesting, and almost every article on this topic gets it wrong.

In October 2011, Bing’s Duane Forrester said publicly that meta keywords is one of roughly a thousand signals Bing analyzes, then clarified what kind of signal it is. It isn’t a positive ranking input. It’s an integrity check. His reasoning, paraphrased closely: if you’re willing to stuff pointless keywords into a tag nobody ranks on, what else are you willing to do? Three years later, on the Bing Webmaster Blog, Forrester wrote that the tag was “pretty clear… dead in terms of SEO value”.

Read those two together and the asymmetry is the point. On Google, a stuffed keywords tag is inert. On Bing, it’s evidence.

Bing’s index also feeds DuckDuckGo, Yahoo, and Ecosia, and it is what Copilot and ChatGPT’s web search lean on. A tag you consider harmless is being read by more of your traffic than you think.

Google’s news-specific variant went the same way, quieter. The news_keywords tag launched for Google News publishers in 2012, and support was dropped around October 2017 without an announcement. John Mueller confirmed it only when asked directly: support went away when the tag came out of the help center, and keeping it on pages is fine, Google just doesn’t use it for Google News anymore. If your theme still emits it, it’s decoration.

Only Yandex still claims to read them

Yandex documents the tag in its own webmaster help. In the list of meta tags Yandex takes into account, keywords is described as something that can influence a page’s relevance to search queries. That’s a vendor statement about relevance, not a promise of ranking weight, and the page hasn’t materially changed in years.

Here’s the honest state of the field in 2026:

Search engine	Uses the keywords tag?	What that means for you
Google	No	Confirmed publicly since September 2009
Bing	Yes, as a spam signal	Stuffing it can count against you
DuckDuckGo, Yahoo, Ecosia	No signal of their own	They inherit Bing’s judgment, including the spam read
Brave Search	No	Independent index, no documented use
Yandex	Documented as a relevance input	Worth filling only if Russian-language search is a real channel
Baidu	Contested, never documented	Claims run both ways with no primary source; treat as unproven
AI search and assistants	No	Retrieval is embedding-based; the tag isn’t part of the pipeline

One engine says it helps, one says it flags you, and the rest don’t care. That’s not a tactic. That’s a coin flip with a downside.

You’ll see confident claims that Naver and Seznam use the tag internally too. Unsourced. Leave those out of your planning until somebody points at documentation.

The plugins already decided this for you

If you’re hunting for a meta keywords field and can’t find it, nothing is broken. The plugin removed it on purpose.

Rank Math doesn’t output the tag at all by default. What Rank Math gives you is a Focus Keyword field, and that field is a content-analysis input, not HTML. It feeds the on-page score, the suggestions, and the schema hints. It never reaches your page source. Getting them confused is the single most common misreading of Rank Math’s interface.

Yoast SEO removed the feature outright. Yoast’s own explainer states that since Yoast SEO 6.3 the feature ceased to exist, and there is no longer any option to switch a keywords field back on. Yoast’s stated reason lines up with Bing’s: misused, the tag reads as a spam signal rather than a ranking signal.

All in One SEO kept the capability and buried it. In AIOSEO 4, meta keywords moved to Search Appearance, under the Advanced tab, as a Use Meta Keywords toggle that ships off. AIOSEO’s own documentation explains the demotion plainly: the field has no value for the majority of users, and the one search engine that still uses meta keywords is Yandex.

SEOPress and The SEO Framework sit on the same side of the line, with SEOPress leaning toward exposing the control for people who ask and The SEO Framework leaning toward not shipping settings it considers dead weight.

Four independent vendors, competing hard against each other, arrived at the same conclusion within a few releases of one another. Settled.

If you’re weighing the two biggest options on other grounds, I compared them in detail in Rank Math vs Yoast.

Reading a competitor’s tag still pays

Writing the tag is worthless. Reading it is occasionally excellent.

A competitor who still fills in meta keywords has handed you their internal keyword list. Not the terms they rank for, which any tool will tell you. The terms they want, in their own words, including the ones they haven’t built pages for yet. That gap between declared intent and published content is the most useful thing on the page.

What most people do with a competitor page:

Run the URL through a rank tracker, export the ranking keywords, add the ones with volume to a spreadsheet.

What works better:

View source on the page, pull the keywords tag if there is one, then subtract everything they already rank for. What’s left is their stated roadmap.

Same page, two minutes apart. The first tells you where a competitor is today, which every tool already told you. The second tells you what they think they should own and haven’t claimed yet, because a keywords tag is written by whoever briefed the page, not by the SERP.

This works on a shrinking share of the web, and it works best in exactly the industries that still run old plugin stacks: local services, regional e-commerce, B2B suppliers, older publishers. If you compete with anyone in that set, it’s worth the look.

Check what your site is still shipping

The tag being ignored by Google doesn’t mean the tag on your site is safe. It means the tag on your site is unsupervised, which is a different condition.

Legacy keyword values survive plugin migrations routinely. They sit in the database, a theme function or an old option keeps printing them, and nobody notices because nobody views source on their own posts. A site that switched from All in One SEO Pack to Rank Math in 2019 can still be emitting a fifteen-term keywords tag written by a contractor a decade ago.

Start with one command against any published URL:

curl -s https://yoursite.com/your-post/ | grep -i 'name="keywords"'

Silence is a pass. A comma-separated list you don’t recognize means something in your stack is still writing the tag, and that’s the thing to fix.

If it does print something, find the stored values before you go hunting through theme files. The legacy meta keys depend on which plugin wrote them and when:

wp db query "SELECT meta_key, COUNT(*) FROM wp_postmeta
WHERE meta_key IN ('_aioseop_keywords','_yoast_wpseo_metakeywords','_yoast_wpseo_metakey')
AND meta_value != '' GROUP BY meta_key"

A count in the hundreds tells you the values are real, still stored, and worth deciding about deliberately instead of leaving to a plugin you may replace next year.

Adjust the table prefix if yours isn’t wp_, and read before you write. A count query is safe; a delete query against postmeta is not something to run without a backup.

What to do with what you find is a judgment call, and it’s a small one. Removing historical keyword values from old posts is low-value cleanup unless the values are genuinely stuffed or off-topic, in which case Bing’s spam read is the reason to bother. Turning off whatever is printing the tag going forward costs a minute and closes the question permanently.

The documented way to add the tag

There is one clean reason to want the keywords tag on a WordPress site in 2026: Russian-language search through Yandex is a channel you’re actively working. That’s it. If that’s you, use the route your plugin documents rather than pasting a meta tag into a header template.

Rank Math needs a filter, not a settings toggle

There is no checkbox for this. Rank Math exposes the behavior through a filter, documented in its own knowledge base:

add_filter( 'rank_math/frontend/show_keywords', '__return_true' );

That prints the post’s focus keyword as the keywords tag, sitewide, from the moment it runs. It’s the whole implementation.

Rank Math documents the file as rank-math.php in your active theme folder. Put it in a child theme or a small site-specific plugin instead, or a parent-theme update will silently delete it. Any code snippets plugin works as well.

If you searched for a Rank Math setting called Custom Meta Tags under Titles & Meta and couldn’t find it, that’s because it doesn’t exist. Instructions describing that path are wrong, and following them wastes an afternoon.

All in One SEO buries a real toggle

AIOSEO is the one plugin here with a supported interface for this. Go to Search Appearance, open the Advanced tab, and set Use Meta Keywords to Yes. Per-post keyword fields become available afterward.

If your only reason for choosing an SEO plugin is Yandex keyword output, AIOSEO is the shortest path to it. That is a narrow reason to pick a plugin.

Yoast needs your own snippet

Yoast won’t do this for you at any price, so you write it. The version worth using pulls from WordPress post tags, which you’re already maintaining:

add_action( 'wp_head', function () {
    if ( ! is_singular( 'post' ) ) {
        return;
    }
    $tags = get_the_tags();
    if ( empty( $tags ) || is_wp_error( $tags ) ) {
        return;
    }
    $names = wp_list_pluck( $tags, 'name' );
    printf(
        '<meta name="keywords" content="%s">' . "\n",
        esc_attr( implode( ', ', $names ) )
    );
}, 5 );

Sourcing from tags rather than a manual field is the part that matters. A field decays the moment someone forgets it; tags stay accurate because they’re doing another job on the site already, and a tag list nobody stuffs can’t become the spam signal Bing is watching for.

Meta keywords vs focus keyword vs target keyword

Most of the confusion around WordPress meta keywords isn’t about the tag. It’s about three terms that sound identical and live in completely different places. Only one of them ever appears in HTML.

• Meta keywords is an HTML tag in the page head. Output, public, ignored by Google, watched by Bing.
• Focus keyword is a plugin setting in Rank Math, Yoast, SEOPress, or AIOSEO. It never reaches your page source. It exists so the plugin can check whether the term appears in your title, URL, first paragraph, headings, and image alt text.
• Target keyword is a strategy decision that exists whether or not any plugin is installed. It’s the query you built the page to win.

The practical order for a new post runs target keyword first, focus keyword second, meta keywords never. Pick the query from research, set it as the focus keyword so the plugin can grade the alignment, write the page to satisfy the query, and leave the keywords field empty.

The field you should care about is the one that changes nothing on the page and everything about how you check your own work.

Where this advice stops working

Leaving the field blank is the right default, and it settles less than it sounds like.

It can’t tell you whether Bing has already read a stuffed tag on your archive, because there is no report anywhere that surfaces meta keywords as a factor. Bing Webmaster Tools won’t show you a spam-signal score. So if your site spent years emitting off-topic keyword lists, cleaning them up gives you no measurable before-and-after; you’re removing a liability, not booking a gain, and you should expect a flat graph.

It can’t settle Baidu. Chinese search has no primary documentation on this question in either direction, and the secondary sources contradict each other confidently. If Baidu is a real channel for you, that’s a question for someone doing Baidu work weekly, not for an article about a WordPress field.

It won’t survive a plugin migration on its own either. Turning off keyword output in one plugin doesn’t delete the stored values, and the next plugin may find them and start printing them again. The audit query above is the thing to re-run after any SEO plugin change, which is the part you still have to remember yourself.

And it changes nothing about ranking. Nobody has ever recovered a page by clearing a keywords tag. This is hygiene, not a lever.

What quietly makes this worse

None of these look like mistakes while you’re making them.

Filling the keywords field because the plugin displays it. An empty input reads as unfinished work, so people finish it. The cost isn’t the thirty seconds; it’s that a field with content in it looks maintained, so nobody ever audits it again.

Copying your competitor’s keywords tag into your own. The logic feels sound, since they clearly researched it. What you’ve actually done is import someone else’s brand terms onto your pages, which is the exact construction Bing’s spam read was built to catch.

Treating the Rank Math focus keyword as a keywords tag. It isn’t one, and the mistake usually runs the other way: people set a focus keyword, assume it’s now public metadata, and stop checking whether the term appears in the copy where it counts.

Adding the tag through a header template instead of the documented route. It works until the theme updates, and then it either vanishes silently or duplicates once the plugin also starts emitting one. Two keywords tags on a page is worse than none.

Reading a 2015 tutorial and assuming the plugin still works that way. Yoast removed the feature in 6.3, AIOSEO moved it behind an advanced toggle, Rank Math never shipped it as a setting. Any instruction set that predates those changes will send you looking for a control that isn’t there.

Spend the time on scored signals

The five minutes you just got back is a real, if small, budget. Here’s where it earns something, roughly in order of return per hour:

1. The title tag. Still the highest-return string on the page, and the one most often left as an auto-generated duplicate of the H1.
2. The meta description. Not a ranking factor, and that’s beside the point. It’s the ad copy for a listing you already earned, and click-through gains compound at a fixed position.
3. Internal links. A handful of contextual links per article, with anchor text that describes the destination rather than the sentence it sits in.
4. Entity precision. Named tools, versions, prices, dates, and standards, in place of category nouns. This is what determines whether an AI answer engine can cite you, and I’ve written up the full pass in Entity SEO Audit.
5. Schema markup. Article, HowTo, Product, and Review types, mostly generated for you by whichever plugin you already run.
6. Core Web Vitals. LCP under 2.5s, INP under 200ms, CLS under 0.1. This is hosting, caching, and image handling, not plugin configuration.
7. Refreshing older posts. Current data, current screenshots, current tool names. Nothing else on this list moves an aging article as reliably.

Every item on that list is scored by something. Meta keywords are scored by nobody, except as a hint that a site might be cutting corners.

If your reason for asking about meta keywords was really “am I visible in AI answers,” that’s a different question with a different answer, and the tag has nothing to do with it. Start with the AI search content optimization checklist, then work through the GEO playbook if you want the strategy layer.

The field was never the work

The reason meta keywords refuse to die isn’t that anyone believes in them. It’s that they look like SEO. A labeled box, a list of terms, a save button, and a feeling of having optimized something. Actual search work in 2026 looks like judgment: which query, which angle, which evidence, which page gets built and which gets merged. None of that fits in a form field, which is exactly why the form field survived.

So the trade is honest but lopsided. You give up a task that felt productive and you get back a small amount of time, plus one less unsupervised string in your page head. Nobody’s rankings move either way.

Run the curl command once, on any post. Then leave the field alone for good.