Titles cut off: check by pixels, not by characters

The client sends a screenshot of his own Google search: the title breaks off in the middle of the brand name, the description ends before the phone number. Snippets like that often sit in the result list that way for months, because nobody googles their own site and Search Console shows no excerpt, only clicks, impressions and a click-through rate that comes out just a little lower than that of the neighbors on the same position. The rank does not suffer. The click-through rate does.

How to go about it in JMX

1. Crawl the site from the Crawl tab. JMX records the title and meta description per HTML page in characters and in pixels; it measures Arial at the size Google sets the result list in — 20 pixels for the title, 14 pixels for the snippet.
2. Open Results → Page titles and show the Title length and Title pixels columns. Sorted descending by pixels, the work list is right there: from 580 pixels Google truncates on the desktop.
3. The same under Results → Meta descriptions with Description length and Description pixels. Here the limit is 985 pixels.
4. In the Issues tab the same pages appear grouped by rule. SEO.TITLE.TOO_WIDE and SEO.META_DESCRIPTION.TOO_WIDE report as a warning and name per page how many pixels too many there are; SEO.TITLE.TOO_LONG (over 60 characters) and SEO.META_DESCRIPTION.TOO_LONG (over 155 characters) sit next to them as notices — the character count is the rule of thumb while writing, the pixel width is the decision.
5. Click an affected row and look in the detail area under Preview → Google snippet: same width, same number of lines. What ends in an ellipsis there ends that way in search too.
6. The By segment breakdown in the Issues view says whether shop, blog or careers is affected. Who writes the texts depends on that.
7. The Tasks tab works out the rest: these rules carry the effort “small” per affected page, so 300 truncated titles are 300 texts and not one template change. The list goes to the editorial team as PDF, Excel or CSV.

What to watch out for

The most common misreading: 580 pixels is not a promise. JMX measures the HTML as served, not the result page — Google regularly rewrites title links itself, with text from headings, anchor texts or the domain name, and does so even for titles that sit comfortably under the limit. Conversely, a truncated title is not a ranking problem, and the description is not a ranking factor at all. What is gone is only the part that was supposed to make someone click.

The width is an approximation, and the finding therefore says “about”. The calculation uses the advance widths of Arial, and it is done for the desktop — mobile result lists wrap differently, and JMX has no measurement for that. Whether the truncation actually costs clicks is answered not by the crawl but by Search insights under “Good position, almost no clicks”. The thresholds themselves are defaults and not constants of nature: title length, title pixels and description length can be changed under Settings → Thresholds, and a re-evaluation needs no new crawl.

One special case is regularly taken for a bug: pages with noindex do not appear in these findings. That is deliberate, because every one of these rules argues from the search result, and a page that never appears in a result has no truncated snippet — otherwise a filtered shop navigation with a few hundred hits would bury exactly the pages this is about. External addresses, redirects and error pages stay out for the same reason.

Anyone who has sorted the Title pixels column once writes differently afterwards: the load-bearing terms up front, the brand name at the end — not because that is more elegant, but because only the beginning is guaranteed to be read.