Content efficiency
Agents prioritise readable text, so a page that is mostly markup costs them more to read.
- Category
- Agent experience
- Standard
- Recommended
What it checks
The extension fetches the page’s HTML and measures how much of it is readable
text. It strips <script> and <style> blocks and every tag, then divides the
length of the text that is left by the length of the whole HTML response. The
check passes when text makes up at least 5% of the HTML.
Below that, markup dominates: an agent pays for thousands of bytes of wrappers, inline scripts and hydration data to get at a small amount of content.
Results
| Status | When |
|---|---|
| Pass | Readable text is 5% or more of the HTML |
| Warn | Readable text is less than 5% of the HTML |
| Unmeasured | The page’s HTML could not be fetched |
How to fix
Reduce the markup around your content rather than adding more text:
- Move large inline
<script>and<style>blocks into external files. - Server-render the content instead of shipping it inside a large JSON hydration blob alongside an empty shell.
- Keep wrapper nesting shallow; ten nested
<div>elements around a paragraph add bytes and no meaning.
A related check, Page token budget, looks at the opposite problem: a page with so much text that an agent cannot read all of it.