Script coverage (used vs unused)

Of the JavaScript shipped, how much actually ran during load (used vs unused bytes, Chrome coverage).

LCP INP CLS

At a glance the headline numbers for Script coverage (used vs unused)

Of the JavaScript shipped, how much actually ran during load (used vs unused bytes, Chrome coverage).

1.9s

Median INP stack

the measures at p50

54%

Biggest: Used kb

1.0s of the stack

1.7s

Over budget at p50

vs the 200ms target

866 KB of JavaScript on the typical page never ran during load.

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv

Where the milliseconds go the phases at the median, against the budget

200ms target

Used kb 1.0s · 54% Unused kb 866ms · 46%

At the median the whole stack uses 1.9s - already past the 200ms INP budget. The biggest piece is Used kb at 1.0s, 54% of the total. computed

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv

Passing LCP per phase the share of sites passing, by how long each phase takes

Used kb Unused kb

Each line is one phase: left = sites where it is quick, right = sites where it drags (buckets, short to long). The steeper the fall, the more that phase decides LCP.

Used kb falls hardest: from its fastest to its slowest bucket the LCP pass rate drops 18 points, down to 72%. computed

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv

The 2 distributions how each phase spreads across sites - color is the share passing in that range

Used kb p50 1.0 MB · p75 1.9 MB

0–4993000+

Unused kb p50 866 KB · p75 1.5 MB

0–4992500+

mostly passing tipping mostly failing

Used kb stays green across its whole range: it never decides LCP. Used kb turns red in its tail - the failing sites live there. computed

The State of Web Vitals · Q2 2026 · 189,915 sites · phone field datacorewebvitals.io/state-of-cwv

Why this matters for the Core Web Vitals, and where to start fixing it

Chrome's coverage data splits the JavaScript you ship into bytes that ran during load and bytes that did not. The unused half still cost the full price: downloaded, parsed and compiled on the main thread, all before doing nothing.

Unused JavaScript is usually whole features shipped everywhere: the checkout bundle on the blog, admin widgets for anonymous visitors. Code splitting per route is the structural fix. Deleting dead dependencies is the cheap one.

Go fix it Reduce unused JavaScript → 14 methods to defer JavaScript → Defer scripts until needed →

Related signals Image source mix → Page weight & requests → DOM depth → Scripts per page → Chrome field data from 189,915 sites, representing millions of real page loads · How we measured

Live queries (4) — admin only

Query #1: 287.8 ms

SELECT COUNT(*) AS count,
            quantile_disc(m.scripts.coverage.used_kb, 0.10) AS p10,
            quantile_disc(m.scripts.coverage.used_kb, 0.25) AS p25,
            quantile_disc(m.scripts.coverage.used_kb, 0.50) AS p50,
            quantile_disc(m.scripts.coverage.used_kb, 0.75) AS p75,
            quantile_disc(m.scripts.coverage.used_kb, 0.90) AS p90,
            quantile_disc(m.scripts.coverage.used_kb, 0.99) AS p99
            FROM sites WHERE m.scripts.coverage.used_kb IS NOT NULL;

Query #2: 300.7 ms

SELECT CASE WHEN m.scripts.coverage.used_kb >= 0 AND m.scripts.coverage.used_kb < 500 THEN 0 WHEN m.scripts.coverage.used_kb >= 500 AND m.scripts.coverage.used_kb < 1000 THEN 1 WHEN m.scripts.coverage.used_kb >= 1000 AND m.scripts.coverage.used_kb < 1500 THEN 2 WHEN m.scripts.coverage.used_kb >= 1500 AND m.scripts.coverage.used_kb < 2000 THEN 3 WHEN m.scripts.coverage.used_kb >= 2000 AND m.scripts.coverage.used_kb < 2500 THEN 4 WHEN m.scripts.coverage.used_kb >= 2500 AND m.scripts.coverage.used_kb < 3000 THEN 5 WHEN m.scripts.coverage.used_kb >= 3000 THEN 6 END AS bucket_idx, COUNT(*) AS n,
            quantile_disc(crux."phone".lcp, 0.5) AS median,
            COUNT(*) FILTER (WHERE crux."phone".lcp IS NOT NULL AND crux."phone".lcp <= 2500) * 1.0 / NULLIF(COUNT(*) FILTER (WHERE crux."phone".lcp IS NOT NULL), 0) AS good_pct
            FROM sites WHERE m.scripts.coverage.used_kb IS NOT NULL GROUP BY bucket_idx;

Query #3: 275.9 ms

SELECT COUNT(*) AS count,
            quantile_disc(m.scripts.coverage.unused_kb, 0.10) AS p10,
            quantile_disc(m.scripts.coverage.unused_kb, 0.25) AS p25,
            quantile_disc(m.scripts.coverage.unused_kb, 0.50) AS p50,
            quantile_disc(m.scripts.coverage.unused_kb, 0.75) AS p75,
            quantile_disc(m.scripts.coverage.unused_kb, 0.90) AS p90,
            quantile_disc(m.scripts.coverage.unused_kb, 0.99) AS p99
            FROM sites WHERE m.scripts.coverage.unused_kb IS NOT NULL;

Query #4: 276.1 ms

SELECT CASE WHEN m.scripts.coverage.unused_kb >= 0 AND m.scripts.coverage.unused_kb < 500 THEN 0 WHEN m.scripts.coverage.unused_kb >= 500 AND m.scripts.coverage.unused_kb < 1000 THEN 1 WHEN m.scripts.coverage.unused_kb >= 1000 AND m.scripts.coverage.unused_kb < 1500 THEN 2 WHEN m.scripts.coverage.unused_kb >= 1500 AND m.scripts.coverage.unused_kb < 2000 THEN 3 WHEN m.scripts.coverage.unused_kb >= 2000 AND m.scripts.coverage.unused_kb < 2500 THEN 4 WHEN m.scripts.coverage.unused_kb >= 2500 THEN 5 END AS bucket_idx, COUNT(*) AS n,
            quantile_disc(crux."phone".lcp, 0.5) AS median,
            COUNT(*) FILTER (WHERE crux."phone".lcp IS NOT NULL AND crux."phone".lcp <= 2500) * 1.0 / NULLIF(COUNT(*) FILTER (WHERE crux."phone".lcp IS NOT NULL), 0) AS good_pct
            FROM sites WHERE m.scripts.coverage.unused_kb IS NOT NULL GROUP BY bucket_idx;

JSON file lookups (4)

                    20260625/meta.json
                    0.03 ms
                

                    20260625/types.json
                    0.56 ms
                

                    20260625/loaf-scripts.json
                    0.46 ms
                

                    20260625/menu.json
                    0.13 ms
                