Reading reports

Every report is ordered to surface what matters first, and to hide the rest until you ask for it.

6 min read

Executive brief

The field verdict, the lab score and the top risk, in that order. Written to be read by someone who will never open the rest of the page — a stakeholder, a client, your future self in three months.

Critical and needs improvement

The urgent issues and the recommended fixes, ranked by measured cost to your visitors. When your site passes every vital, the Critical section is hidden rather than filled — false urgency is how a report loses its credibility.

What a finding contains

Each one names the problem, quotes what was measured on your page (the request, the element, the byte count), states the fix, and estimates the recovery. Where a finding can be verified after a deploy, it says how.

What the order actually means

The fix list is not sorted by severity label or by audit ID. It is ordered by what each finding is measured to cost your visitors — the savings Lighthouse attributes to it, weighted by whether the field data shows that metric failing for real users.

That is why a small image fix routinely outranks a large architectural one, and why the order can change between two scans of the same page: if your LCP recovers, the things that were blocking it stop being expensive.

Each finding also carries an effort tag. Impact and effort are separate axes on purpose — the top item is the best return, not the easiest job.

What you can safely skip

Needs improvement, when Critical is populated. Fix the top of the list first; the lower items often resolve themselves.
Lab opportunities with a green field verdict. Real users are fine. These are housekeeping, not urgency.
Anything with savings under ~100 ms. Below the noise floor of what a visitor can perceive, and below what a re-scan can reliably confirm.

When a page passes every vital, the Critical section is hidden rather than filled with make-work. An empty Critical section is a result, not a rendering fault.

Technical diagnostics

The full detail behind the summary: lab opportunities, the CrUX distribution for each metric, and the trend over your own scan history. This is where you go to answer “why”, not “what”.

Any report can be shared as a point-in-time link or exported as text or PDF, so the person reading it sees the same numbers you did rather than a fresh scan with different ones.