Initial stylesheet transfers exceed the recorded budget

Why it matters

Stylesheets can delay the browser's first rendering while it reads the rules needed to display a page. Reducing unnecessary CSS can help, but the transfer count alone does not establish which rules are critical or how quickly visitors see the page.

What to do

1. Read the recorded page, stylesheet addresses, encoded transfer sizes and budget. A missing or incomplete source is stated separately; it is not counted as proof of a complete page.
2. Remove unused rules and unnecessary stylesheet links, and consider splitting CSS by page template. Keep the styles the page needs.
3. Use compression where appropriate and compare the result in a browser. Inline only a suitably small critical portion when you have established what the page needs; this check does not make that judgment.

When to leave it as it is: You can leave the stylesheets as they are when their size is intentional and acceptable for your visitors. The audit reports a budget comparison, not proof of a poor experience.

What we read

We use the page's original HTML head and its base address to identify direct stylesheet links confirmed to apply to screen media at the start of the page. Disabled, inert, body-only and print-only links do not establish a blocking transfer. Unresolved media conditions, named stylesheet sets and uncertain head placement leave coverage incomplete.

For each confirmed requested address, we count the encoded response-body bytes actually measured by our reader. Content-Length and decoded CSS size are not substitutes. Repeated links to one captured requested address count once; separate captures at different requested addresses count separately even when they reach the same final address.

Imported child stylesheets are outside this direct-link budget. An incomplete transfer can establish a positive lower bound when the bytes already exceed the budget, but cannot establish that a previous problem was fixed. The budget comes from the recorded audit settings; an older audit without it remains unchecked.

Example & evidence

Illustrative example — schematic recorded counters, not a live audit.

A page at https://bakery.example/ links two stylesheets in its initial head for screen media. The recorded budget is 102,400 bytes.

/assets/base.css: 61440 encoded bytes
/assets/page.css: 40961 encoded bytes
Total: 102401 bytes
Recorded budget: 102400 bytes

The measured total exceeds the budget by one byte. If a new complete audit reads the same page and measures 61440 + 40960 = 102400 bytes, the total no longer exceeds the budget. An incomplete transfer or unreadable page cannot confirm that correction.

Confirm & check again

Check for yourself: Open the browser's Network panel and inspect the stylesheets, their transferred sizes and loading conditions. Browser cache, compression and selected media can differ from our recorded requests.

Check again: Full audit — Run a new full audit. A correction is confirmed only when the same page and all relevant sources have sufficient coverage and their total no longer exceeds the recorded budget. Missing or incomplete evidence leaves it unconfirmed.

Limits & sources

We do not render the page, identify critical CSS or measure a browser waterfall, first paint or Largest Contentful Paint. Work limits, unavailable sources and unresolved conditions leave coverage incomplete. The measured transfer count describes our requests, not every possible visitor's network or cache.

Confidence: High The comparison uses recorded counters and a recorded budget; its effect on a visitor's loading experience is not measured.

Sources

• web.dev, Optimize Largest Contentful Paint — Explains how render-blocking resources can delay rendering and discusses reducing unused styles and resource load time. The product's 100 KiB budget is not a limit specified by this guide. Checked 10 October 2026.

Editorial review: 15 October 2026