Your hreflang is wrong: a technical SEO audit checklist for SaaS

40 audits. The same errors, ranked by frequency:

1. Missing self-referencing hreflang (the page forgets to include itself, classic).
2. Wrong language codes (nl-BE written as be, which is Belarusian, congratulations on your Minsk traffic).
3. Hreflang pointing at redirected URLs (the tags rot while the site moves).
4. No x-default, or x-default pointing at a language picker nobody uses.
5. Blocking the alternate URLs in robots.txt while declaring them. Declaring and hiding, simultaneously.

Fix order: screaming frog or equivalent crawl, extract all hreflang, validate codes against the actual registry, recrawl after every deploy. Two hours, once a quarter. Helsinki precision, I accept payment in coffee.

Content-side confirmation: the be/Belarusian error is my favourite bug in all of SEO. Second favourite: hreflang clusters where every page declares a different canonical. The tags agree, the canonicals feud.

PHP forever, multilingual invoicing forever: our cs/sk/en triangle broke quarterly until we added hreflang validation to CI. Tests for SEO tags felt absurd. Absurd and effective.

CI validation for hreflang is advanced and correct. Absurd is deploying untested tags to production. You would not deploy untested code. Tags are code.

Onboarding-email parallel: locale bugs in emails are the same shape. Wrong segment, wrong language, sent with confidence. Validate the segment, not just the sentence.