A good SEO title states, in the space of one line, exactly what this page contains and for whom — using the words a searcher would use — and a good meta description expands that promise into two short lines that make the right reader choose this result. Write them per page, not by template formula: lead with the page's specific topic, keep titles near the roughly 50–60 characters that usually display without truncation and descriptions near 120–160, make every one unique across your site, and never promise in the tag what the page does not deliver. Our SEO Checker shows the title and description of any URL with their character counts, so you can check the tags as served rather than as drafted.

Start with the measurements, not the folklore

Rules of thumb about length are repeated so often they harden into law. The honest version: Google truncates by pixel width, not character count, varies the space by device, and may rewrite both the title link and the snippet from your page content when it judges its version a better match for the query. Character counts are a practical proxy, nothing more. To ground this guide, we fetched five live pages while writing it and measured the tags exactly as served — title and description text extracted from the HTML and counted character by character:

Page Title (measured) Chars Description (measured) Chars
example.com "Example Domain" 14 Missing 0
MDN, HTTP status codes reference "HTTP response status codes - HTTP | MDN" 39 "HTTP response status codes indicate whether a specific HTTP request has been successfully completed…" 139
RFC 9110 (HTTP Semantics) "RFC 9110: HTTP Semantics" 24 Missing 0
schema.org, Article type "Article - Schema.org Type" 25 (not captured in this pass) —
Google Search Central, Article structured data docs "Learn About Article Schema Markup | Google Search Central | Documentation | Google for Developers" 97 Present (not quoted here) —

Three lessons fall out of real data. First, short is not automatically wrong: the RFC's 24-character title names the document exactly, and no searcher needs more. Second, missing is a genuine loss: example.com and the RFC both leave the snippet entirely to the search engine's improvisation, because there is no description to draw on. Third, long is a real risk: Google's own documentation title, at 97 characters, is far beyond what a results line displays — most of it exists for the browser tab and the page itself, not the result. Aim for the MDN row: a title that names the topic and the source, and a description that is itself the first sentence of the answer.

Write the title: topic first, brand last

The working structure for most pages is Specific topic – useful qualifier | Brand. The topic comes first because truncation eats the end, and because the first words are what a scanning searcher reads. The qualifier earns its space only if it adds a distinction a searcher cares about — a year for time-sensitive material, a format ("checklist", "calculator"), an audience ("for landlords"). The brand goes last, or is dropped on your most competitive pages where space is tight; your site name already appears in the result's URL line.

Match the words to the intent behind the query, in the searcher's vocabulary, not your internal naming. A page your team calls "Onboarding resources" that answers "how to set up email on a new domain" should be titled for the second phrase. One page, one intent: a title trying to cover three topics covers none of them distinctly, and usually signals the page itself needs splitting — the duplicate failure dissected in How to Find and Fix Duplicate Titles and Meta Descriptions.

Write the description: the honest trailer

The description is not a ranking input; Google says so, and it may replace yours with text lifted from the page. Write it anyway, because when it is used, it is your one chance to speak to the searcher directly, and a page with no description hands that chance to chance. Two sentences usually suffice: what the page gives the reader, and what makes this version worth choosing — a worked example, tested data, a decision table. Use the active voice, include the page's key phrase naturally where it fits, and keep every claim true of the page: a description that over-promises produces a visit that bounces in seconds, which helps no one. If the page's first paragraph already summarises it well, the description can be a tightened version of it — several strong sites do exactly this, as the MDN row above shows: its description is the definition its title promises.

Worked rewrites: before and after

The "before" values below are the measured ones from the table; the "after" values are rewrites we wrote for this guide, labelled as proposals — they are editorial examples, not captures of pages that changed.

Page Before After (proposed) What changed
example.com Title: "Example Domain" (14 chars). Description: missing. Title: "Example Domain – Reserved for Use in Documentation" (50 chars). Description: "Example domains are reserved for documentation and examples. They may be used in illustrative examples without prior permission or coordination." (144 chars) The rewrite says what the domain is for, which is the only reason anyone lands there; the description states the terms of use a visitor actually needs.
RFC 9110 Title: "RFC 9110: HTTP Semantics" (24 chars). Description: missing. Title kept — it names a standard precisely. Description: "RFC 9110 defines HTTP semantics: methods, status codes, header fields and message framing for the core web protocol, independent of any one implementation." (155 chars) A case where the title needs nothing and the fix is only the missing summary — restraint is part of the craft.

Notice the discipline: neither rewrite adds keywords the page cannot support, neither uses clickbait the content cannot cash, and both stay inside the display space that the measurements show is realistic.

A repeatable workflow

  1. List the page's intent in one sentence before writing anything: "This page helps [who] do [what]." If you cannot write the sentence, the page has the problem, not the tag.
  2. Draft the title from that sentence, topic first, then count it. Over ~60 characters, cut qualifiers before cutting the topic.
  3. Draft the description as the trailer for the page as it exists today — not the page you intend to improve later.
  4. Check uniqueness against the rest of the site; identical or near-identical tags go through the cluster process in the duplicate-metadata guide.
  5. Verify as served. Fetch the live page or run it through the SEO Checker and read the tags from the output. Templates, SEO plugins and caches can all serve something other than what you saved; the checker run we used for the audit guide's worked example reported title, description length and the issues "Missing meta description" and "Title is very short" straight from the fetched HTML of example.com — exactly the verification this step needs. Tag problems found this way across a whole site belong in the routine sweep of How to Perform a Website SEO Audit Step by Step.

Common mistakes

  • Keyword-stuffing the title with every variant of the phrase. It reads as spam to searchers, truncates badly, and invites Google to rewrite the title link entirely.
  • One template description for the whole site. A description that fits every page describes none; if you cannot write a unique one yet, leaving the tag off a page is better than duplicating one — Google will excerpt the page itself.
  • Writing tags for the page you wish you had. Every claim in a tag is a promise; the page must keep it in the first screenful.
  • Chasing a character counter as if it were a rule. Counts are proxies for pixel width; a 58-character title of wide characters can truncate while a 65-character narrow one displays fully. Judge the draft as a line of text, not a number.
  • Rewriting every tag at once after a traffic dip. You lose the ability to tell what helped. Change tags in small batches, note the date, and compare like with like.

Frequently asked questions

How long should a title tag be?

There is no character limit in HTML and no fixed display width. In practice, titles around 50–60 characters usually display in full on desktop results; longer titles still work — the words beyond the visible part are read even when not shown — but front-load what matters so truncation never removes the topic. The measured 97-character documentation title in this guide's table shows what happens without that discipline.

Does Google always use my title and description?

No. Google may generate the title link from your headings or other text, and it builds the snippet from your description or from page content, choosing per query. A clear, accurate title tag and a faithful H1 that agree with each other give Google the least reason to improvise; agreement between tag, heading and content is the controllable part.

Are meta descriptions a ranking factor?

Google states they are not used as a ranking signal. Their value is click-through: a result whose snippet matches the searcher's need precisely gets chosen over a generic one. That is an indirect benefit, and it is the honest reason to write them well.

Should every page have a unique description?

Every page you want in results should have a description written for it. For large catalogues, generate descriptions from each page's real data (its name, category, key attribute) rather than one shared sentence — generated-from-data is unique and factual; copied boilerplate is neither.

How often should titles be revised?

When the page changes intent or content, when you find the search results for your target query have shifted to a different intent, or when a title is demonstrably mismatched — not on a schedule. Titles are part of the page's identity; churn without cause makes your own results history unreadable.

References

Related reading: Structured Data Generator, canonical tags explained.