The WordPress Speed Optimization Checklist We Run on Every Audit

Speed optimization fails when it happens in random order: teams tweak scripts while the site still lacks a page cache, or buy CDNs while shipping 600KB of unused CSS. This checklist is ordered by dependency — each step assumes the ones before it — and it is the sequence our audits follow. Roughly 30 checks, grouped in six phases.

Phase 1: Measure honestly (before touching anything)

• Baseline Core Web Vitals from field data (Search Console) — the thresholds in the CWV guide.
• Lab waterfall (WebPageTest, mid-tier mobile) — record TTFB, LCP element, total weight.
• Test cached and uncached URLs separately — caches flatter numbers and hide the real tax.

Phase 2: Caching (the foundation)

• Page cache present and exclusive? One layer only — plugin or server-level, never both (the layer conflicts in the caching guide).
• Cached TTFB under 200ms? If not, the cache is not actually serving.
• Exclusions correct? Cart, checkout, account, member pages — verify with logged-in tests.
• Cache purges on publish? Edit a post, check the front end immediately.

Phase 3: Images (usually the biggest LCP win)

• All images WebP/AVIF? Count legacy JPEG/PNG in uploads.
• LCP image eager and preloaded? Lazy-loaded hero images are the classic self-inflicted LCP wound (documented in the optimizer comparison).
• Images sized to their containers? A 2000px file in a 400px slot is wasted bytes on every view.
• Dimensions set? Missing width/height attributes cause CLS — free fix.

Phase 4: Theme weight (the structural tax)

• CSS coverage: what share of shipped CSS does the page use? Under 60% means theme bloat — measure with the method in the coverage audit.
• Uncached page weight: under ~350KB total is a healthy target for content pages.
• DOM size: builders routinely produce 2,000+ node pages; each node costs parse time.
• Uncached LCP: if it stays past 3s after this phase, the theme is the bottleneck — see the switch framework.

Phase 5: Scripts and fonts

• Third-party script inventory: chat widgets, analytics, trackers — each with weight and whether it blocks (the pattern in the third-party guide).
• Scripts the page does not need unloaded per-template — the tooling in the script manager comparison.
• Fonts: count files loaded vs used; preload one; self-host — the font guide.

Phase 6: Server and data layer

• PHP version current? Each major PHP release is measurably faster.
• Object cache available and connected? Redis/Memcached for logged-in and uncached paths — setup here.
• Autoloaded options under 800KB? The every-request tax described in the DB test.
• wp-cron on real cron? Visitor-triggered cron adds latency to random page loads.
• CDN for static assets if the audience is international — measured honestly in the CDN comparison.

The report format that drives fixes

Each finding gets three fields: the metric, the cost (in ms or KB), and the fix owner. That turns a 30-item list into a ranked backlog — usually five items cover 80% of the gap. The free audit runs this exact sequence and returns the ranked version.