The script is 40 lines. It takes a sitemap, fetches every URL one at a time, pulls <title>, and prints a length. I ran it against a flashcard site we run and got 160 pages over 60 characters.
The real number is 144. Both errors were mine, and both are the kind that make a data post useless if you don't catch them.
Error one: entity length
& is one character of title and five characters of source. Twenty pages in that set have an ampersand in the book title. I counted the source. Fix: html.unescape() before len().
Sixteen pages flipped. That's 8% of the site, from one missing function call.
Error two: threads
The first version used a thread pool of eight. It returned 28 errors on a 191-page site and I spent twenty minutes looking for missing pages that weren't missing — they were timeouts that surfaced as failed requests, indistinguishable from a real negative answer.
The rewrite is single-threaded with a 250 ms sleep. 191 pages, 48 seconds, zero errors, and a 404 now looks like a 404.
What the 191 pages actually look like
| Page type | Pages | Mean title | Shortest | Longest | Over 60 chars |
|---|---|---|---|---|---|
| Book-card pages | 156 | 83 chars | 52 | 136 | 133 |
| Blog posts | 23 | 58 chars | 44 | 64 | 10 |
| Landing, legal, index | 11 | 50 chars | 26 | 64 | 1 |
| All pages | 191 | 78 chars | 26 | 136 | 144 |
Every URL returned 200 on the first attempt, so there's no dead-link noise in this dataset.
The part I expected to be bad and wasn't
I assumed the head tags were a mess. They aren't. Zero pages are missing a description. Zero of the 191 descriptions collide with another. Every canonical points at its own page. Exactly one H1 per page, no stray robots meta. Nine of ten checks pass.
One field is broken and it's the one nobody looks at, because the checker tool most people use shows you the title length and you think "I'll get to it."
The boilerplate tax
All 156 book pages end with the identical string:
Flashcards: Key Concepts & Review | WakeCards
Forty-six characters, on every page, before you write anything about the book. Which leaves fourteen characters of budget. The median book title on the site is 37.
Strip the brand suffix and the mean drops 78 → 66, and pages over 60 drop from 144 to 98. So half of it is a suffix, and half of it is that we wrote a sentence where a noun phrase belonged.
What the script can't tell you
Whether Google truncates any of these. Truncation is pixel-width at a font size I don't set and can't query. Google also rewrites titles it doesn't like, and the rate is a Search Console number I didn't pull. Sixty characters is a convention from tooling, not a fact about Google.
What I can measure is the input: 144 pages carry 25.8 characters more than that convention allows.
Disclosure
The site is wakecards.com, one of ours. The script and the page list are in this pass only — no sampling, all 191 URLs.
The fix is 156 rewrites with 156 different tails. No template. That's the part that doesn't fit in 40 lines.

