Resources & Tools
Case Studies: How Hosting Choices Impacted Real Businesses (AU)
By Dominik Kowalski · Updated 21 Aug 2025 · 12–16 min read
Purpose: Show measurable, real‑world outcomes from targeted hosting improvements—what changed, how we measured it, and what you can replicate.
TL;DR: AU‑first audiences often benefit from an AU origin + CDN, strong caching, and image optimisation. Expect faster LCP and lower INP in field data—and, when done right, tangible conversion lifts.
Case A: Local clinic (AU origin + caching)
Industry Healthcare (AU) · Stack WordPress · Audience 95% AU · Goal More online bookings
Baseline (field + lab)
| Metric (mobile, AU) | Before | Source |
|---|---|---|
| TTFB | 1100–1300 ms | WebPageTest Sydney |
| LCP (P75) | 3.8 s | CrUX |
| INP (P75) | 280 ms | CrUX |
| Bookings conversion rate | 2.4% | Analytics (annotated) |
Changes
- Migrated origin to Sydney region; enabled page caching and Redis object cache.
- Added CDN with AU POPs; served hero images as AVIF with responsive sizes.
- Preconnected fonts; inlined critical CSS for above‑the‑fold content.
After (4 weeks post‑change)
| Metric | After | Delta |
|---|---|---|
| TTFB | 480–650 ms | −40–55% |
| LCP (P75) | 2.2 s | −1.6 s |
| INP (P75) | 160 ms | −120 ms |
| Bookings conversion rate | 2.8–3.1% | +16–29% |
What we’d do differently: Earlier audit of third‑party scripts (booking widget). Related: AU hosting & SEO
Case B: Niche ecommerce (caching + media discipline)
Industry Retail (AU/NZ) · Stack WooCommerce · Audience 80% AU · Goal More revenue per session
Baseline
| Metric | Before | Source |
|---|---|---|
| LCP (P75) | 3.1 s | CrUX |
| INP (P75) | 340 ms | CrUX |
| Revenue per session | Indexed 100 | Analytics |
Changes
- Edge caching for category and product pages; strict cache‑control for static assets.
- Automated AVIF/WebP pipeline; dimensioned images to reduce CLS.
- Deferred non‑critical JS (reviews, chat); capped total JS on PDPs.
After (30 days)
| Metric | After | Delta |
|---|---|---|
| LCP (P75) | 2.3 s | −0.8 s |
| INP (P75) | 190 ms | −150 ms |
| Revenue per session | Indexed 109 | +9% |
What we’d do differently: Earlier route split to keep checkout JS lean. Related: Speed & conversions
Case C: SaaS marketing site (multi‑region/edge)
Industry B2B SaaS (global) · Stack Next.js + edge functions · Audience AU/EU/US · Goal Stable launches under load
Baseline risks
- Single‑region origin causing high TTFB for distant users during campaigns.
- Spikes led to queueing and occasional 5xx during product launches.
Changes
- Adopted edge rendering for marketing pages; multi‑region origins for APIs.
- Introduced streaming responses and smarter cache keys.
- Added load shedding and circuit breakers for third‑party scripts.
After (launch week)
| Metric | Before | After |
|---|---|---|
| Global 5xx rate (peak hour) | 0.9% | <0.1% |
| AU TTFB (median) | 950 ms | 420 ms |
| Landing page LCP (P75) | 3.0 s | 2.1 s |
What we’d do differently: Pre‑warm edge caches for hero campaigns. Related: Shared vs VPS vs Cloud
How to reproduce (AU checklist)
- Measure baseline: CrUX by country, WebPageTest Sydney, analytics funnels.
- Fix images: AVIF/WebP, responsive sizes, set dimensions to avoid CLS.
- Cache what you can: HTML for anonymous traffic, static assets long‑lived.
- Reduce JS: defer non‑critical, cap bundle size on money pages.
- Origin strategy: AU origin for AU‑first sites; add CDN with AU POPs.
- Re‑measure: compare comparable periods, annotate changes in analytics.
See also: Top hosting mistakes and AU hosting & SEO.
Need a second opinion? For pragmatic audits and migration plans, visit Dominik Digital Marketing. Start with a performance assessment or go direct: https://domdigitalmarketing.com.au/.