WordPress Speed Optimization for International Stores: A Core Web Vitals Playbook
A template-by-template playbook for WordPress and WooCommerce Core Web Vitals: LCP, INP, CLS, images, fonts, CSS, JS, caching and international delivery.
Speed advice for WordPress tends to be a list of plugins. This playbook is different: it works from the metric that is failing, identifies the template responsible, and fixes the actual cost.
It is written for international ecommerce stores, where mobile traffic dominates and latency to a single origin can be brutal.
Know which metric you are fixing
| Metric | What it measures | Typical ecommerce culprit |
|---|---|---|
| LCP | When the main content finishes rendering | Hero image, render-blocking CSS, slow TTFB |
| INP | Responsiveness to interaction | Heavy JavaScript, third-party tags, long tasks |
| CLS | Visual stability | Images without dimensions, late-loading fonts and ads |
| TTFB | Server response time | Slow hosting, uncached queries, no object cache |
Start with field data from real users — CrUX in Search Console or your own RUM — then reproduce on the templates that carry revenue: home, category, product, cart, checkout.
Set budgets per template
Targets that work for mid-range mobile devices:
- LCP: under 2.5 seconds; under 1.8 seconds on product pages if possible.
- INP: under 200 milliseconds.
- CLS: under 0.1.
- Page weight: 1.5 MB or less for a product page; under 1 MB for text pages.
- Requests: under 60 on first load; every third-party script must justify itself.
1. Fix the image pipeline
Images are usually the largest LCP cost. Serve modern formats (AVIF or WebP), correctly sized for the layout, with explicit width and height. Preload the hero image on the templates that use one; lazy-load everything below the fold, but never the LCP element.
In WordPress, check that your theme is not generating 2560-pixel images for a 600-pixel card, and that a compression plugin is not re-encoding on every request.
2. Treat fonts as a design budget
Two families and four or five weights is plenty. Self-host, subset to the languages you actually serve, preload the one font used above the fold, and use font-display: swap. A third-party font CDN adds a DNS lookup and a privacy question in every market.
3. Ship critical CSS, defer the rest
A page builder often loads its entire UI kit on every template. Extract the above-the-fold CSS, inline it, and defer the remainder. Remove unused component CSS when you migrate templates rather than layering overrides forever.
4. Audit JavaScript, especially third parties
Open the network panel and sort by size. Then ask, for each script: does this help a buyer complete an order? Chat widgets, heat maps, A/B tests and consent tools can each cost more than the feature is worth. Load them after interaction or on a delay, and remove anything nobody has opened in three months.
5. Cache with the cart in mind
Page caching works only if cart, checkout and account pages are excluded correctly. Pair it with object caching (Redis) for product queries and tax rules. Test as a logged-in customer, as a guest with items in the cart, and as a returning customer with a session.
6. Fix the origin and the edge
No front-end trick fixes a two-second TTFB. Use hosting with adequate PHP workers, current PHP 8.x, and a database that is not held together by autoloaded options. For multi-region audiences, add a CDN and consider edge caching for anonymous traffic.
7. WooCommerce-specific costs
- Product queries that load every variation on archive pages.
- Cart fragments scripts loading on every page for every visitor.
- Faceted navigation generating slow, duplicate URLs.
- Related-product queries without limits.
- Plugins that inject their assets sitewide instead of on the pages that need them.
A 10-day speed sprint
- Days 1–2: measure field and lab data per template; agree targets.
- Days 3–5: fix images, fonts and critical CSS.
- Days 6–7: audit and defer JavaScript; remove dead tags.
- Days 8–9: caching, database and hosting tuning.
- Day 10: re-test, document before/after, set up ongoing monitoring.
Three common mistakes
Chasing a perfect lab score. A 100 Lighthouse score with a broken mobile menu is worse than a 78 with a fast checkout. Prioritise revenue templates.
Installing five optimisation plugins. They conflict, each adds overhead, and you cannot tell which one helped. Choose one caching layer, one asset optimisation layer, and fix the rest in code.
Optimising only the homepage. Most organic and paid traffic lands on product and category pages. Measure those.
If you want the diagnosis without the trial and error, our speed sprints start at $900 and include field-data verification. Not sure whether speed is your real problem? Request the free 20-point audit — we will tell you if the bottleneck is elsewhere.