Performance Improvements
- Tackling LCP 7.6s
Recorded a mobile performance score of 67 on PageSpeed Insights. LCP (Largest Contentful Paint) was slow at 7.6s, with element render delay at 3,100ms. We reduced initial animations, optimized images, and revised the GTM loading strategy. Meanwhile, Day 6 AEO measurements showed Citation 0 and Mention 3, regressing from Day 5.
AEO Measurement Results Summary (Day 6)
0 / 14
← Day 5: 1/140% - Regressed from Day 5
3 / 14
← Day 5: 5/1421% - Decreased from Day 5
| Metric | #001 (2/1) | #002 (2/2) | #003 (2/3) | #004 (2/4) | #005 (2/5) | #006 (2/6) |
|---|---|---|---|---|---|---|
| Citation | 0/14 (0%) | 0/14 (0%) | 0/14 (0%) | 0/14 (0%) | 1/14 (7%) | 0/14 (0%) |
| Mention | 4/14 (29%) | 3/14 (21%) | 3/14 (21%) | 3/14 (21%) | 5/14 (36%) | 3/14 (21%) |
"AEO tool" query: On Day 5, the official LP (sighted.tkbase.co.jp) was cited with mention: true, citation: true. On Day 6, this dropped to mention: false, citation: false. The AI response format changed to distinguish between two contexts of AEO: logistics/customs (Authorized Economic Operator) and web marketing (Answer Engine Optimization), and Sighted was not mentioned in either.
On Day 5, Sighted was cited via a PR TIMES press release, but the citation source may have changed on Day 6. The addition of structured data and performance improvements may have had an effect, but direct causation is unclear.
• "What is Sighted's AEO scoring pricing plan?" - mention: true, citation: false (Same as Day 5. Effect of structured data addition to be checked in next measurement)
• "How to register for Sighted's beta on the official site?" - mention: true, citation: false (Same as Day 5. Confusion with sightedapparel.com continues)
• "How to improve AI search visibility with Sighted?" - mention: true, citation: false (Same as Day 5. Other services like sightedby.ai being cited instead)
PageSpeed Insights Results (Before Improvements)
67
Mobile (target: 90+)
7.6s
Element render delay: 3,100ms
• LCP 7.6s: Element render delay 3,100ms (caused by HeroSection initial animation)
• Image optimization: dashboard-demo.png 508.8 KiB → 494.7 KiB reducible (Next.js Image optimization disabled)
• Render blocking CSS: 150ms blocked
• JavaScript execution time: 1.5s reducible (including GTM)
• Unused JavaScript: 203 KiB reducible (including GTM 107.4 KiB)
• Forced reflow: Multiple occurrences related to framer-motion
• Legacy JavaScript: 11 KiB of unnecessary polyfills
Improvements Implemented
Problem: HeroSection was wrapped in a Reveal component with initial opacity: 0, causing 3,100ms element render delay.
Fix: Removed the Reveal wrapper from HeroSection in app/page.tsx for immediate display. Also removed the motion.div animation from dashboard-demo.png, replacing it with a standard div.
Expected impact: Element render delay 3,100ms → 0ms (significant LCP improvement)
Problem: images.unoptimized: true was set in next.config.mjs, disabling image optimization. dashboard-demo.png (3830x2028px) was being served at full size despite displaying at 637x337px.
Fix: Updated next.config.mjs to enable Image optimization with AVIF/WebP format support and configured deviceSizes/imageSizes. Added sizes attributes to dashboard-demo.png and sighted-logo-full.png.
Expected impact: dashboard-demo.png 494.7 KiB reduction, sighted-logo-full.png 7.1 KiB reduction
Problem: GTM Script was loaded with strategy="afterInteractive", blocking initial rendering. 107.4 KiB of unused JavaScript was reducible.
Fix: Changed GTM Script in app/layout.tsx to strategy="lazyOnload", deferring load until after page load completes.
Expected impact: Reduced initial JavaScript execution time and unused JavaScript
Problem: The official site was not being cited for queries like "What is Sighted's pricing plan?" and "How to register for the beta?" The content answering these questions was not machine-readable.
Fix: Added SoftwareApplication offers for Standard (4,890 JPY/month) and Enterprise (contact us) plans. Added FAQPage entries for pricing and beta registration. Also explicitly distinguished from Sighted Apparel in the FAQ.
Expected impact: AI can now accurately extract pricing and beta information when citing. To be verified in next measurement.
Improvement Details
• app/page.tsx: Removed HeroSection Reveal wrapper
• components/landing/hero-section.tsx: Removed motion.div animation from dashboard-demo.png, added sizes attribute
• next.config.mjs: Enabled Image optimization, AVIF/WebP support
• app/layout.tsx: Changed GTM Script loading strategy to lazyOnload, expanded productSchema offers (pricing structured data)
• components/ui/logo.tsx: Added sizes attribute to sighted-logo-full.png
• components/landing/structured-data.tsx: Added pricing and beta FAQs to FAQPage
• components/landing/faq-section.tsx: Added pricing and beta questions to display FAQ
• LCP: 7.6s → ~3-4s (3,100ms element render delay eliminated)
• Image size: ~500 KiB reduction (auto-resize and format conversion via Next.js Image optimization)
• JavaScript execution time: Initial execution reduced (GTM lazy loading)
• Performance score: 67 → ~75-80 (to be verified in next measurement)
AEO Measurement Results Analysis
Q: Why did the Citation for "AEO tool" disappear?
On Day 5, Sighted was cited via a PR TIMES press release, but on Day 6 the AI response format changed, splitting the explanation between logistics AEO and web marketing AEO. As a result, Sighted was no longer mentioned. Since AI responses vary each time, this could be a temporary fluctuation, though the effect of structured data additions and performance improvements cannot be ruled out.
Q: Why did Mentions drop from 5 to 3?
Mentions were lost for two queries: "AEO tool" and "AEO monitoring." For "AEO tool," the cause was the response format change described above. For "AEO monitoring," a Best Calendar article was cited on Day 5 but not on Day 6. This demonstrates the inherent variability in AI responses.
Q: What effect did the structured data addition have?
Pricing and beta structured data (Schema.org / FAQPage) was added, but no effect was observed in the Day 6 measurement. Structured data may take time to produce results. Verification is needed in subsequent measurements (Day 7 onwards).
Q: How do performance improvements affect AEO?
LCP improvements and image optimization may affect the speed at which AI crawlers load the site. However, direct causation with the Citation disappearance on Day 6 is unclear. AI response variability involves multiple factors (training data updates, citation source changes, query interpretation changes, etc.), requiring continuous monitoring.
Next Actions
- Verify improvement impact in next PageSpeed Insights measurement (LCP, performance score)
- Verify structured data addition effect (check Citation/Mention changes for pricing and beta queries in Day 7+ measurements)
- Aim to regain citation for "AEO tool" query (strengthen awareness via PR TIMES or enhance official LP content)
- Improve render-blocking CSS (consider async CSS loading)
- Resolve forced reflows (review framer-motion usage)
With Sighted, you can run the same kind of analysis shown in this article for your own site. Measure Citations and Mentions, benchmark against competitors, and verify the impact of your optimization efforts.