Performance-First WordPress Build

Context

A clean slate, built right from the first line.

A tourism operator in Patagonia needed a professional web presence built from scratch. No existing site to migrate, no legacy code to work around. A clean slate — and an opportunity to build it right from the first line.

Problem

Two seconds to earn trust, on mobile.

Tourism is a mobile-first, trust-first industry. Visitors arrive via Instagram, Google, or a referral link — almost always on mobile. If the site doesn't load fast and look credible in the first two seconds, the booking doesn't happen.

The challenge wasn't fixing a broken site. It was building one that wouldn't need fixing later.

Diagnosis

The regional standard left room to win.

Pre-build audit of the niche: most tourism WordPress sites in the region were running GTmetrix grades of C or D, with LCP above 3 seconds and page weights above 5MB. The gap between standard and excellent was achievable with disciplined build decisions.

Constraints

What the build had to work within.

• Budget-appropriate stack — no premium CDN or enterprise hosting
• Client non-technical — site had to be maintainable without developer dependency
• Launch deadline — no time for iterative optimization post-launch

Implementation

Performance decided before a single page went live.

Stack selected for performance before design: Astra Pro as the theme base for its minimal footprint, WP Rocket configured before any content was published, Cloudflare proxy active from day one, Wordfence hardened at launch.

No demo imports. Every page built from blank canvas. Plugin count kept to the functional minimum — each plugin installed has a specific job and nothing else.

Images optimized before upload. Lazy loading configured. Fonts loaded with font-display: swap. No render-blocking scripts.

Tradeoffs

Slower to build, cleaner to maintain.

Building from scratch without demo imports takes longer than importing a template and modifying it. The tradeoff is zero legacy CSS, zero unused JavaScript, and a DOM that stays clean as the client adds content over time.

Stack

Before / After

New build — baseline established at launch, not after.

New build — performance baseline established at launch, not after.

GTmetrix Grade A · Performance 92% · LCP 1.1s · May 2026

SSL Labs Grade A · May 2026

PageSpeed Mobile 97 · Best Practices 100 · SEO 100 · May 2026

PageSpeed Desktop 98 · LCP 0.6s · May 2026

Outcome

Launched on deadline, no optimization sprint needed.

Site launched within deadline, Core Web Vitals in the green, GTmetrix Grade A. Client can update content independently without breaking performance. No optimization sprint required post-launch.

Lessons

Performance is a build decision, not a fix.

Performance is a build decision, not a post-launch fix. The difference between a GTmetrix A and a GTmetrix C is almost entirely in the decisions made before the first page goes live — stack selection, plugin discipline, and image workflow.