Meta Description Length Checker – Characters, Pixels and the Real Cut Point
The meta descriptionis the only sentence of a page that most people will ever read. It is not a ranking factor, and it never has been, but it is the sales copy of the search result — and it is the one piece of a page's metadata that gets written to a character count that does not decide anything. Search engines lay a description out in a fixed-width column and cut it when the line runs out of room, which means the truncation happens on rendered width, not on a character total. This meta description length checker measures both, so the number you write to is the number that actually applies.
Why 155 characters is the wrong target
Character guidance survives because it is easy to count, not because it is correct. At 13-pixel Arial a lowercase i advances 2.89 pixels and a capital W advances 12.27 — over four times as far. A description of 155 narrow characters fits inside the desktop budget with room to spare; the same 155 characters set in capitals overflows it badly. The tool sums a published per-character advance-width table across your text and compares the result against roughly 920 pixels on desktop and 680 on mobile, then reports the worse of the character verdict and the pixel verdict. Because most search traffic is mobile, the narrower budget is usually the one that matters.
What the tool actually checks
Before anything is counted, the text is normalized the way a browser would render it: HTML tags stripped, entities such as & decoded, non-breaking spaces and newlines turned back into ordinary spaces, and runs of whitespace collapsed. Those artefacts arrive with almost every copy-paste out of a CMS and quietly inflate a raw count. What remains is measured for characters, UTF-16 units, UTF-8 bytes for database field limits, words, sentences and estimated pixel width.
On top of the length work it runs the checks that quietly cost clicks: whether your focus keyword appears and whether it lands inside the first 60 characters where the bolding does the most good, whether a call to action is present, how much of the page title the description merely repeats, and whether the copy carries straight quotes, emoji, smart quotes or stray tabs that will misbehave inside an HTML attribute. Those feed a quality score out of 100 whose points are shown individually, so a low score always names its cause.
Previews, bulk audits and raw HTML
The centrepiece is a live search result preview that renders your title, URL and description at desktop and mobile widths and applies the ellipsis at the real pixel boundary, dimming the words that would be lost rather than hiding them. When the description overruns, the rewrite assistant offers trimmed variants — a clean word-boundary cut, one with stock filler phrases removed first, and one with whole trailing sentences dropped so the copy still reads as finished.
Raw HTML mode takes a pasted <head> and pulls the description, og:description and twitter:description out of it, flagging where they disagree. Bulk mode takes a crawl export — one description per line, or a URL and a description separated by a tab — and returns a scored table with per-row status, duplicate detection across rows, and CSV export. That is how a site audit gets done in one pass instead of one page at a time.
What is estimated, and what is exact
Character counts, byte sizes, word counts and keyword positions are exact. Pixel widths are estimates: the advance-width tables are correct for Arial and Roboto, but real rendering also involves kerning, hinting, the font installed on the reader's device and subpixel rounding, so read the figure as accurate to within a few percent. Accented letters are charged their base letter's width, which is right because a diacritic adds height and not width, while non-Latin characters are charged a full em — correct for full-width scripts and deliberately conservative elsewhere. Everything runs in your browser, so an unpublished page's metadata never leaves the machine.