Meta tag generator: build the full head block and take it ready to paste

Who this page is for: anyone who needs the head tags written and checked in one pass, instead of hunting each one down in a different tutorial. If what you want is to see where Google cuts your text, measured in pixels, the place is the SERP snippet preview. If what you want is structured data markup, which is a different format with a different job, the path is the Article schema generator or the FAQ schema generator. And if the subject is the file that controls crawling across the whole site rather than the tags of one page, see the robots.txt generator with AI bots. The scope here is one thing only: the tags that live inside the head of a page.

The block has four layers, and almost every site forgets one

Meta tags are not a loose list of optional lines. They are four groups with different audiences, and most sites fill in the first, forget the second and have never heard of the third:

• The rendering basics. The charset tag tells the browser which encoding the file uses, and without it accented characters turn into symbols. The viewport tells it the page adapts to the screen width, and without it phones draw the site at desktop size and the visitor has to pinch to read.
• What search reads. The title tag, the meta description, the canonical and meta robots. This is the group almost everyone remembers, and even here the canonical tends to be missing, which leaves the same page reachable through different addresses competing with each other.
• What social reads. The Open Graph tags, which apply to Facebook, LinkedIn, WhatsApp, Slack, Discord and virtually every app that draws a link preview. Without them your shared link shows up as a raw address, next to competitors with a title, a description and an image.
• What X reads. The card tags, which decide whether the link becomes a large image or a thumbnail beside the text. They fall back to Open Graph, but the card format itself is only set by its own tag.

The generator assembles all four groups at once, in the order they should appear, and leaves out everything that no longer does anything: meta keywords, revisit-after, distribution and the whole fossil list that old generators still emit. An empty field does not become an empty tag, it simply does not show up.

How to use it

1. Write the title and the description of the page. Both fields are measured in characters and in pixels as you type, on the same Arial ruler the SERP preview uses.
2. Paste the full canonical URL, with https and no campaign parameter. It is the address you want in search when the same page exists at more than one path.
3. Fill in the social image and its alt text. That pair is what separates a link with a good looking preview from a link that looks broken on WhatsApp.
4. Set the switches: index and follow stay on for a normal page; charset and viewport can be turned off if your page already declares them.
5. Check the score and the warnings list, hit copy block and paste it inside the head of your page, before it closes.

How it works: the scoring ruler

The score runs from 0 to 100 and is the sum of eleven weighted checks. Each one matches a known way the block fails in the real world, and the list below is the whole ruler, with nothing hidden:

The title and description checks award half credit when you get one measure right and the other wrong, because the two measure different things. The character count is the convention everyone verifies; the pixel count is the one Google actually uses, since it draws the text and cuts when the space runs out. A 58 character title full of wide letters overflows the line, and a 64 character one with narrow letters passes in full. The pixel ruler is the same one used by the SERP snippet preview, on purpose: two different rulers would give two different numbers for the same text.

One detail almost no generator handles: field content is escaped before it becomes an attribute. Quotes, greater than signs and ampersands you type are converted, and line breaks become spaces. Without that, a single quote in the middle of the title closes the attribute early and breaks the HTML of the whole page, which is a quiet and very common way to take down your own social preview.

Worked example (reproduces the default output)

The example already loaded is a course page. Checking the panel, number by number:

• Title: 51 characters and 479 pixels, inside the 600 pixels of one desktop line. The title check reads 16 of 16.
• Description: 146 characters and 865 pixels, inside the 1,200 pixels of the two lines. The check reads 14 of 14.
• Block: 21 tags generated and 1.7 kB in size. That is 2 basics, 6 for search, 8 for Open Graph and 5 for the X card.
• Score: 100, band A, with all eleven checks green and no warnings in the list.

Now swap the title for a longer one with the brand at the end, something like the course text followed by "of live classes with a real project on your own site" and the school name, reaching 102 characters. Without touching anything else, the panel answers: 966 pixels against the 600 available, the title check drops to 0 of 16, the score goes to 84 and the band falls from A to B. Two warnings appear, one about the character count and one about the pixel overflow, which is the measure that decides. The other ten checks stay intact: the score does not collapse, it shows exactly what that one field cost.

How to read each number

The score is the quick read, and band A is the target. It answers whether the block is ready to paste, not whether the page will rank. A perfect block on a weak page is still a weak page with a well written head, and that is exactly why the score is worth having: it clears away the part that is pure execution work.

The two numbers on each text field should be read together. The character one is the market convention and what your client will check; the pixel one is what decides the cut. When the two disagree, believe the pixels and explain the difference to whoever is asking for 60 characters.

The tag count is the fastest way to notice that a field came out empty. The complete block ships 21 tags. If yours reads 17, four fields are blank, and the checks list says which.

The warnings do not deduct points on their own: they explain the deduction the checks already made, and some show up even at a full score, like the one telling you the X handle was normalized to the at sign format. It pays to read them all before copying.

Known limits

The tool checks format, not existence. It looks at whether the canonical is an absolute address and whether the image is on https, but it does not visit the URL or download the file, because that would require a network call and break the promise that nothing leaves your browser. An image with a perfect address and a deleted file passes the score and fails in real life: after publishing, check the preview in each network validator.

On the search side, the usual warning applies: Google rewrites the title in a good share of queries and the description in most of them. A well written block does not force anyone to use it, it removes the known reasons for the swap, and a cut in the middle of a sentence is one of the most common.

The reference widths are the desktop ones, with a 600 pixel line. On mobile the space is different, the title gets a second line and the cut moves. The tool uses the stricter measure on purpose, so text approved here fits both formats, and anyone who wants to see both drawn has the SERP preview for that.

Finally, the block covers head tags, not structured data. Review stars, prices and the questions that appear expanded on search come from JSON-LD, which is a different format living in a different tag. If you want those, build the schema too with the FAQPage generator and paste both blocks on the same page.

From the pasted head to the tested click

Writing the tag is the beginning. Once it is live, the question becomes which version makes more people click and what happens to the ones who do:

• See your snippet drawn and the exact cut point in the SERP snippet preview, and check the mobile format too.
• If the destination is a landing page, run it through the landing page grader before fighting for more clicks to a page that does not convert.
• Before deciding whether one version of the page beats another, check how much traffic the test needs in the sample size calculator.
• With the test running, read the verdict in the statistical significance calculator instead of calling it on the first good day.
• To know whether the site is ready to be cited by AI systems, not just ranked, use the GEO readiness checklist.

Frequently asked questions

Which meta tags are actually required?

Three: the title tag, the meta description and the charset tag. The title is what Google draws as the blue text of the result and what social networks use as the card name. The description is the argument that wins the click. The charset keeps accented characters from rendering as garbage. The rest of the block is not required in the technical sense, but every missing field is a decision you left to a machine: with no og:image your link becomes plain text on WhatsApp, and with no canonical two identical URLs compete against each other.

Does meta keywords still do anything?

No, which is why it is not here. Google stopped using meta keywords as a ranking signal back in 2009 and said so publicly. No relevant search engine reads that tag today. A generator that still emits meta keywords is copying a 2008 template: the practical effect of keeping it is handing your keyword list to anyone who opens your page source. The same goes for meta revisit-after and meta distribution, which never did anything in any search engine.

Do I need to repeat the title inside og:title and twitter:title?

You do need to declare them, and the generator declares them for you. Networks read og:title first, and only fall back to the title tag when og:title is missing. Repeating the same text in all three places is the default behavior and it is what this tool does. If you want different wording for social, which often pays off when the search title is very descriptive and social calls for something punchier, edit the block after copying: the separate fields exist in HTML precisely to allow that difference.

What is the right size for the og:image?

The reference that works across every network at once is 1200 by 630 pixels, at a 1.91 to 1 ratio, under 5 MB and served over https. LinkedIn and Facebook crop to that ratio; X uses the same one in the large card format. An image narrower than 600 pixels is usually rejected and the card falls back to text. One common trap: the image URL has to be absolute, starting with https, because the social crawler reads the tag outside your page context and has no way to resolve a relative path.

Does the block score guarantee the page will rank?

No, and no honest tool would promise that. The score measures what is under your control: the block is complete, the addresses are absolute, the title fits the result line, the description fits the two lines and does not repeat the title. Ranking depends on content, authority and competition, which no tag solves. What a good block does is remove reasons for the engine to ignore what you wrote and rewrite the snippet on its own, and make sure a shared link shows up with an image instead of showing up naked.

Where exactly do I paste this block?

Inside the head tag of the page HTML, before it closes, and only once per page. On WordPress, an SEO plugin usually writes these tags, and having both sources at once produces duplicated tags: in that case use the plugin fields instead of pasting the block. On a hand built site, on Astro, Next or Hugo, the block goes in the head template and the values become per page variables. The classic mistake is pasting it into the body, where no search engine reads it.

Is the robots noindex here the same as the robots.txt file?

It is not, and confusing the two is one of the most expensive technical SEO mistakes. Meta robots with noindex tells the engine it may read the page but may not list it. The robots.txt file tells it not to fetch the page at all. If you block the URL in robots.txt, the engine will never read the meta noindex, and the page can keep showing up in search with no description at all. To pull a page out of the index, leave crawling open and use meta noindex, which is the switch in this tool.

Are the fields I type stored anywhere?

No. Everything runs in JavaScript inside your browser: the block is assembled and the score is computed on your own machine, with no network call, and nothing is stored or logged. The tool also does not visit the canonical URL or download the image you entered, which means it checks the address format and not whether the file exists. That matters when the page is not published yet and the title gives away a launch. You can confirm it by opening the network tab and typing: no request goes out.

Keep reading

To understand how search and AI answers treat what you publish, follow with CRO vs SEO, what changes in each, read the guide to GEO in 2026 and learn to build citable blocks in FAQ content for AI citation.