Fix guides
Each guide covers one issue Audra checks on every page it crawls: what it is, why it matters, the exact markup that fixes it, and how to find every affected page on your site.
Last updated 2026-07-28
- How to fix a missing meta description A missing or duplicate meta description costs you click-throughs. Here is what a good one looks like, the ideal length, and how to find every page missing one.
- How to fix a missing or poorly written title tag The title tag is the single strongest on-page SEO signal. Here is the right length, the right structure, and how to audit every title on your site at once.
- How to fix a missing or duplicate H1 Every page needs exactly one H1 that states what the page is about. Here is why multiple H1s cause problems and how to audit headings across a whole site.
- How to fix images without alt text Missing alt text breaks screen readers and loses image search traffic. Here is how to write it, when to leave it empty, and how to find every offending image.
- How to fix a missing or incorrect canonical tag Canonical tags tell search engines which URL is the real one. Here is how to set them correctly and the mistakes that quietly deindex pages.
- How to fix hreflang errors Hreflang tells Google which language version to show which user. Here are the return-tag, code and self-reference rules that break most implementations.
- How to fix a missing viewport meta tag Without a viewport meta tag, mobile browsers render your site at desktop width and zoom out. One line of HTML fixes it — here is the correct one to use.
- How to fix a robots meta tag that blocks indexing A stray noindex can remove pages from Google overnight. Here is how the robots meta tag works, how it differs from robots.txt, and how to audit it site-wide.
- How to fix a missing html lang attribute The lang attribute tells screen readers which pronunciation to use and helps search engines classify your content. It is one attribute on one element.
- How to fix skipped heading levels Jumping from H2 to H4 breaks screen-reader navigation and muddles your page structure. Here is how to build a heading outline that actually makes sense.
- How to fix thin content Thin content is not about hitting a word count — it is about pages that fail to satisfy the visit. Here is how to identify and fix genuinely thin pages.
- How to fix missing Open Graph tags Without Open Graph tags your links share as a bare URL with no image. Here are the four tags that matter and the image size that works everywhere.
Which of these should you fix first?
Fix anything that stops a page being indexed at all, then anything that affects every page, then everything else. In practice that order is: a stray noindex robots meta tag, a missing or wrong canonical, a missing viewport tag, then title tags and meta descriptions, then structure and content issues.
The reason for that order is leverage rather than severity. A single template-level mistake — one noindex in a theme header, one canonical pointing at the homepage — can suppress hundreds of pages at once, while a weak meta description costs you a fraction of one page's click-through rate. Sorting by "how many pages does this affect" beats sorting by "how bad does this sound" almost every time.
This is also the argument for auditing a whole site rather than a page: the same issue appearing on 40 of 60 pages is a template bug with one fix, and you cannot see that pattern from a single-page test.
What are the exact thresholds Audra flags on?
Audra's HTML checks are deterministic and use fixed thresholds, so a finding is always reproducible. Here are the numbers behind the guides below.
| Check | Flagged when | Severity |
|---|---|---|
| Title tag | Missing, or outside 50–60 characters | Error if missing, otherwise warning |
| Meta description | Missing, or outside 120–158 characters | Warning |
| H1 heading | Zero H1s, or more than one | Error if zero, warning if multiple |
| Heading order | Any skipped level, such as h2 straight to h4 | Warning |
| Canonical | Missing, or not an absolute http(s) URL | Warning |
| Robots meta | Contains noindex, or contains nofollow | Error for noindex, warning for nofollow |
| Viewport meta | Missing | Error |
| html lang | Attribute absent | Warning |
| Image alt text | Any image lacking alt, unless marked decorative | Warning |
| Open Graph | Any of og:title, og:description, og:image, og:url missing | Warning |
| hreflang | Any annotation that is not a valid language tag | Warning |
| Thin content | Under 200 words of body text | Informational |
Frequently asked questions
How do I find every page on my site with one of these issues?
You need a crawler, not a single-page test — the whole point is seeing whether an issue is one page or a template. Audra crawls up to 100 pages by default and lists every URL affected by each finding. Chrome DevTools, Screaming Frog's free 500-URL edition and other crawlers will also do this; the comparison pages cover the trade-offs.
Do these fixes still matter for AI answer engines?
Most of them do, and some matter more than they do for Google. Titles, headings, canonicals and structured data all feed how an answer engine understands and attributes a page, and thin content is quoted far less often. The distinctly AI-specific signals — retrieval-bot access, llms.txt, answer-first structure and freshness — are covered on the comparison pages rather than in these fix guides.