All posts

Turn a Lighthouse Report Into a Plan

July 31, 2026 · DevTools

lighthouse
performance
core-web-vitals
seo
accessibility

A Lighthouse run can produce a hundred individual audit results across four categories. The raw report is thorough, but thorough isn't the same as actionable — faced with that wall of JSON, most people fix one or two obvious things and move on. The trick is to turn the report into a ranked list so you always work on the highest-impact item first.

Start with the four scores

Lighthouse grades four categories from 0 to 100, and the thresholds are consistent:

  • 90–100 — green (good)
  • 50–89 — orange (needs improvement)
  • 0–49 — red (poor)

You don't need the full report to get value — entering just the four category scores gives you the tool's average overview and the weakest area. The Lighthouse Audit Summary does this in its Quick scores mode: type Performance, Accessibility, Best Practices, and SEO, and it tells you which category is dragging the overall number down and flags any red category as high priority. This average is a convenient summary, not an official Lighthouse score.

Then dig into the failed audits

The real wins live in the individual audits, specifically the ones that failed and carry the largest estimated savings in milliseconds. A typical Performance report might show:

  • Render-blocking resources — 1,500 ms
  • Unminified JavaScript — 600 ms
  • Properly size images — 400 ms

Fixing the first one saves more than the other two combined, so it goes to the top of the list. In Paste-report mode, the summary tool extracts zero-score failed audits straight from a Lighthouse JSON report (the file lighthouse --output=json produces) and sorts those findings by estimated savings, so the biggest opportunity in the extracted list is at the top.

Tie Performance to Core Web Vitals

The category score is a lab number; your users live in the field. The three Core Web Vitals are what Google actually measures:

  • LCP (Largest Contentful Paint) — good ≤ 2.5s
  • INP (Interaction to Next Paint) — good ≤ 200ms
  • CLS (Cumulative Layout Shift) — good ≤ 0.1

A page passes only when all three are in the good range, so one weak metric makes the whole assessment poor. Use the Web Vitals Calculator to grade field values, and aim the top of your Lighthouse list at whichever vital is worst.

A repeatable workflow

  1. Quick scores — enter the four numbers to see the overall grade and weakest category.
  2. Paste the report — drop in the JSON to get the failed-audit table ranked by savings.
  3. Fix the top item, re-run Lighthouse, and watch the score move.
  4. Check the field — confirm the Core Web Vitals moved too.

Lighthouse is only useful if it changes what you do next. A ranked list — overall grade, weakest category, failed audits by savings — turns a thorough but overwhelming report into a single "what do I fix first" answer.