How to Prioritize Performance Fixes (Impact Matrix)

After a Lighthouse audit, you're staring at 20+ failures. You can't fix them all at once. The key is knowing which fixes have the highest impact for the least effort.

The Impact-Effort Matrix

Plot each fix on a 2×2 matrix:

HIGH IMPACT
    │
    │  🎯 DO FIRST      ⚡ SCHEDULE
    │  (Quick Wins)     (Worth the effort)
    │
────┼───────────────────────────────
    │
    │  💤 DO IF IDLE     ❌ SKIP
    │  (Low priority)   (Not worth it)
    │
    └──────────────────────── HIGH EFFORT

Prioritized Fix List

🎯 Tier 1: Do First (High Impact, Low Effort)

Fix	Impact	Effort	Lighthouse Points
Compress/resize images	⬆️⬆️⬆️	1-2 hours	+10-25
Add image width/height	⬆️⬆️	30 min	+5-15 (CLS)
Convert to WebP/AVIF	⬆️⬆️⬆️	1-2 hours	+5-15
Remove unused scripts	⬆️⬆️⬆️	1 hour	+5-20
Add font-display: swap	⬆️⬆️	15 min	+3-8
Enable Gzip/Brotli	⬆️⬆️	30 min	+3-8
Preload LCP image	⬆️⬆️	15 min	+3-10

Total effort: ~5-7 hours Expected improvement: +30-80 Lighthouse points

⚡ Tier 2: Schedule (High Impact, Medium Effort)

Fix	Impact	Effort	Lighthouse Points
Code splitting	⬆️⬆️⬆️	4-8 hours	+10-20
Lazy load below-fold images	⬆️⬆️	2-3 hours	+5-10
Inline critical CSS	⬆️⬆️	2-4 hours	+5-10
Defer third-party scripts	⬆️⬆️	2-3 hours	+5-15
Set caching headers	⬆️⬆️	1-2 hours	+3-8
Implement CDN	⬆️⬆️	2-4 hours	+5-10
Remove unused CSS	⬆️	3-5 hours	+3-8

Total effort: ~16-30 hours Expected improvement: +15-40 additional points

💤 Tier 3: Do If Idle (Low Impact, Low Effort)

Fix	Impact	Effort	Lighthouse Points
Optimize CSS selectors	⬆️	1-2 hours	+1-3
Add resource hints (preconnect)	⬆️	30 min	+1-3
Minify HTML	⬆️	30 min	+1-2
Reduce cookie size	⬆️	1 hour	+1-2
Enable HTTP/2 push	⬆️	1 hour	+1-3

❌ Tier 4: Skip (Low Impact, High Effort)

Fix	Why Skip
Rewrite in a new framework	Massive effort for uncertain gains
Custom image CDN	Buy a solution instead
Build custom lazy loading	Use native loading="lazy"
Implement service worker (for performance)	Complex, marginal gains for most sites
Micro-optimize CSS animations	Diminishing returns

Decision Flowchart

When evaluating each fix:

1. Will this fix my worst metric? (Focus on whatever's red first)
2. How many users does this affect? (Fix homepage before admin panel)
3. Can I do this in under 2 hours? (If yes, do it now)
4. Is there a simpler alternative? (Buy vs build, native vs library)
5. Will this compound with other fixes? (CDN + compression = multiplied gains)

Metric-Specific Priorities

If LCP Is Your Worst Metric

1. Preload LCP image/resource
2. Optimize LCP image (format, size)
3. Eliminate render-blocking CSS
4. Reduce TTFB (caching, CDN)

If CLS Is Your Worst Metric

1. Add width/height to all images
2. Reserve space for ads/embeds
3. Preload fonts (prevent FOUT)
4. Avoid inserting content above viewport

If INP/TBT Is Your Worst Metric

1. Remove unused JavaScript
2. Code split heavy bundles
3. Defer third-party scripts
4. Break up long tasks
5. Move work to Web Workers

Track Your Fix Impact

After each fix, re-run Lighthouse to measure improvement. BadPageSpeed tracks your scores over time automatically.

Monitor your improvement →