Title tag length: keep it under 60 characters, and know that the number is a proxy. Google truncates titles by pixel width, roughly 580 pixels on desktop by common observation, which means two titles with the same character count can meet different fates. Ten W's are more than twice as wide as ten i's. The 60-character rule works because titles under it almost always fit; the pixel width is what actually decides.
Meta descriptions follow the same logic at a different size: under 155 characters as the working guideline, truncation by width and layout in practice, and an ellipsis waiting for anything past the budget.
That's the answer. The rest is why the proxy sometimes lies, what Google does to titles it doesn't like, and how to check a snippet before you publish instead of discovering the truncation in the wild.
Why character counts sometimes lie
Search results render in proportional fonts, where every letter has its own width. An i is a sliver, a W is a barge, and a title is a sum of widths, not a count of letters. Google allocates a fixed pixel budget to the title line and cuts what overflows, so the true unit is the pixel.
You can see it with two 58-character titles: one built from narrow letters sails through, one heavy on capitals and wide glyphs loses its last word to the ellipsis. Nothing about the character count warned you. This is why a snippet check that measures actual rendered width beats counting twice, and it's the reason our free SERP preview tool renders your draft in the result layout and reports both numbers: characters for drafting speed, pixels for the verdict.
The same drafting-versus-verdict split applies to descriptions, with one twist: because truncation cuts from the end, a description that answers the query in its first clause keeps its meaning even when trimmed. Answer first isn't just an AI-citation habit; it's truncation insurance.
Does title length affect rankings?
Not the way the worry assumes. Google's search team has said repeatedly that title length itself is not a ranking factor, and that a truncated title is a display event, not an indexing one: the full tag gets read and used for relevance even when the visible version ends in an ellipsis. So a 70-character title isn't penalized. It's just partially invisible, and the invisible part can't persuade anyone.
That reframes what length optimization is actually for. It's click-through work, not ranking work. The title's ranking job (telling Google what the page is about) survives truncation fine; the title's human job (winning the click against nine competitors) dies at the ellipsis if the claim isn't finished by then. Which is also the answer to the perennial brand-suffix question: append "| Your Brand" when the name earns recognition or trust with your searchers and there's room for it, and drop it without guilt when the claim needs the space. The keyword and the claim are load-bearing; the suffix is furniture.
Mobile complicates the numbers without changing the strategy. Mobile results wrap titles onto a second line, so more characters often survive there than on desktop, while descriptions typically show less. Since you can't control which layout a searcher gets, write for the tighter budget and treat anything extra as a bonus.
What Google does to titles it doesn't trust
Google rewrites titles regularly. Not maliciously, and not randomly: when the title doesn't match the query or the page, Google assembles its own from your H1 and on-page text and shows that instead. Clever titles, padded titles, and keyword-stuffed titles get replaced most; plain statements of what the page is get replaced least.
Two practical consequences. First, the title and H1 should agree with each other and with the page, because they're the raw material Google reaches for. Second, a rewrite is feedback, not vandalism: read the version Google chose and you're reading its model of your page. If the model is wrong, the fix is on the page, not in the tag.
We'd also rather flag the limit of the exercise than oversell it: a perfect snippet on a page that doesn't answer anything is a well-dressed vacancy. Snippet work is the last five minutes of a page, never the strategy.
How we draft snippets that survive
The pass we run on every page we rebuild in the Reforge phase of our method, in order:
- Name the one question the page answers. If you can't, stop; no metadata survives an unfocused page. This rule precedes every other rule here.
- Title: keyword early, claim complete, under 60. The end of a title is the part that dies, so nothing essential lives there. Brand suffix only if there's room.
- Description: answer in the first clause, reason-to-click in the second, under 155. Descriptions that open with throat-clearing get rewritten or ignored.
- Check the rendered width, not just the counts, in the SERP preview.
- After indexing, look at the live result. If Google rewrote you, diagnose the gap between your tag and its version.
Here's the same pass as a copyable checklist for your own drafts:
SNIPPET PASS (per page)
[ ] One question this page answers: ______________________
[ ] Title < 60 chars, keyword in first half, claim complete
[ ] Title pixel check passed (no ellipsis in preview)
[ ] Description < 155 chars, answer in first clause
[ ] Title and H1 agree with each other and the page
[ ] Post-index: live SERP checked, rewrite diagnosed if any
Formulas by page type
The budget is universal; what belongs inside it depends on the page. The patterns we reach for:
Homepage: what you do, where, then brand. "AI SEO Agency in Jacksonville | Narsil Creative" tells a stranger and a machine the same three facts in one line. The homepage is the one place the brand earns a permanent seat.
Service pages: service plus city plus the differentiator that fits. "Emergency Plumber in Jacksonville: 60-Minute Response" beats "Plumbing Services" by naming the query and answering the objection in the same breath.
Articles: the question or the claim, stated the way a person would search it. If the piece answers "how long should a title tag be," a title that contains those words in that order is doing the five-placement work already, and cleverness would only tax it.
In every case the test is the same: could a searcher predict the page from the title alone? If yes, the title is done. If it needs the description to explain it, the title is decoration.
Where snippet work pays fastest
Not evenly across the site. The highest-return snippet edits are on pages that already earn impressions without clicks: page one or two rankings where the result shows up and gets skipped. Those are visible in Search Console, and they're first cousins of the striking distance keywords we harvest for quick wins: the ranking work is done, and the snippet is the remaining bottleneck. A title that fits, front-loads the query, and completes its claim is often the cheapest click-through improvement available on those pages.
New sites are the other moment: titles and descriptions are a launch-day item in our website launch checklist, because writing them page by page at launch costs minutes, and retrofitting them across a hundred indexed pages costs a project.
One more place the same text quietly matters: your title and description are also what many platforms fall back to when a page lacks Open Graph tags, so the search snippet you write today may moonlight as tomorrow's share card. Same words, two surfaces, one more reason to write them like they'll be read.
Snippets are small, checkable, and finished in minutes per page, which makes them the rare SEO task with no excuse for guessing. Draft by the numbers, verify by the pixel, and let the page do the ranking.