Technical SEO Experiments
Technical SEO experiments test infrastructure and configuration changes that affect crawl, index, rendering, and page speed.
Learning Focus
After this lesson you can design and run experiments on rendering, crawl paths, page speed, indexation controls, and structured data.
This lesson covers the seven technical experiment types (leaves 10.3.1–10.3.7): rendering changes, internal linking changes, crawl path changes, template changes, page speed improvements, indexation controls, and structured data changes.
Rendering Changes
Test how content rendering approach affects indexation.
Testable rendering elements:
| Element | Test Example |
|---|---|
| SSR vs CSR | Server-side rendered page vs client-side rendered page for the same content |
| JavaScript reduction | Full JS bundle vs code-split bundle |
| Lazy loading | Lazy load below-fold content vs eager load all content |
| Critical CSS | Inline critical CSS vs standard CSS loading |
| Font loading | font-display: swap vs font-display: optional |
Rendering test methodology:
- Implement change on test group.
- Compare rendered HTML (GSC URL Inspection) between test and control.
- Compare indexation rate and crawl behavior.
- Compare CWV metrics.
Internal Linking Changes
Test structural internal linking changes (from Lesson 10.2.5, but at site-wide scale).
Site-wide tests:
| Test | Description |
|---|---|
| Link equity redistribution | Add links from high-authority pages to lower-authority pages |
| Breadcrumb implementation | Add breadcrumbs to all pages vs existing navigation |
| Related content module | Add auto-generated related content links |
| Footer link reduction | Reduce footer links to only important pages |
Crawl Path Changes
Test how changes to crawl paths affect discovery and indexation.
Testable crawl path changes:
| Change | Test Example |
|---|---|
| Sitemap structure | Separate product sitemap vs combined sitemap |
| Robots.txt rules | Block specific parameter patterns vs allow all |
| Internal linking depth | Add links to deep pages vs no change |
| Navigation structure | Simplify top navigation vs current |
Template Changes
Test template-level changes that affect multiple pages.
Template tests:
| Test | Example |
|---|---|
| Canonical implementation | Self-canonical on all pages vs template-level canonical |
| H1 standard | Single H1 vs multiple H1s |
| Schema template | Add automatic schema generation to template |
| Meta tag rules | Change title/description generation logic |
Template test methodology:
- Apply change to a subset of pages (template variant).
- Compare test vs control on: indexation rate, ranking, CTR.
- Run for 2-4 weeks.
- Roll out to all pages if positive.
Page Speed Improvements
Core Concept
Test the SEO impact of page speed improvements.
Testable speed optimizations:
| Optimization | Example |
|---|---|
| Image optimization | WebP format vs JPEG |
| Caching | CDN caching vs no CDN |
| Resource loading | Async JS vs sync JS |
| Font loading | Self-hosted fonts vs Google Fonts |
| Server response | Faster hosting vs current |
Page speed test challenges:
- Page speed improvements are usually site-wide, making page-level A/B testing difficult.
- Use time-based comparison (before/after) with seasonality adjustment.
- Monitor CWV report in GSC for changes in pass rate.
Indexation Controls
Test changes to indexation directives.
Testable indexation changes:
| Change | Test Example |
|---|---|
| Noindex removal | Remove noindex from previously noindexed pages vs keep noindex |
| Canonical addition | Add self-canonical to pages without one |
| Robots.txt change | Allow crawling of previously blocked resources |
| Sitemap submission | Submit sitemap for a URL group vs no sitemap |
Indexation test challenges:
- Indexation changes can take weeks to reflect in GSC.
- Measure the specific pages affected, not site-wide metrics.
- Use GSC URL Inspection to verify indexation state.
Structured Data Changes
Test the impact of adding or changing structured data.
Testable schema changes:
- Add FAQPage schema to existing content vs no schema (only applicable for authoritative government/health sites as of 2024).
- Add Product schema to product pages vs existing schema.
- Change from Microdata to JSON-LD.
- Add/replace Article schema with NewsArticle schema.
- Add missing properties to existing schema.
Schema test measurement:
- Rich result impressions in GSC Enhancements.
- Rich result CTR.
- Organic ranking changes (schema is not a ranking factor, but may affect CTR and eligibility).
Workflow
- Identify the technical change to test (rendering, internal linking, crawl paths, templates, page speed, indexation controls, or structured data). Define a clear hypothesis.
- Apply the change to a defined test group of pages. For site-wide changes (e.g., page speed), use a time-based before/after comparison with seasonality adjustment.
- Validate the change: GSC URL Inspection for rendering, Rich Results Test for schema, GSC Indexing report for indexation state, CWV report for speed.
- Run the test for 2-4 weeks minimum. Compare test vs control on: crawl rate, indexation rate, rankings, and CTR.
- Roll out winning changes in phased stages (5% → 25% → 50% → 100%) with monitoring at each stage.
Common Mistakes
warning
- Not verifying that Google can render the change: Changes to JS-rendered content or lazy-loaded elements may not be seen by Googlebot. Always use GSC URL Inspection to verify the rendered HTML.
- Comparing site-wide changes without seasonality correction: Before/after analysis for site-wide technical changes must account for seasonal traffic patterns. Use year-over-year comparisons.
- Ignoring crawl propagation delay: Technical changes take 1-2 weeks to propagate through Google's crawl. Do not evaluate results before the change is reflected in the index.
- Changing multiple technical elements simultaneously: Template, schema, and page speed changes deployed together cannot be isolated. Deploy technical changes one at a time.
- Not monitoring CWV metrics during speed experiments: Page speed changes can inadvertently hurt CWV scores (e.g., lazy loading can increase CLS). Monitor all three CWV metrics, not just LCP.
Checklist
- Define a clear, testable hypothesis for one technical change
- Select test and control groups (or define time-based measurement plan)
- Verify the change is rendered correctly by Googlebot (GSC URL Inspection)
- Validate schema changes with Rich Results Test
- Collect 4+ weeks of pre-test baseline data
- Run test for minimum 2-4 weeks
- Compare crawl rate, indexation, and ranking differentials
- Document findings and rollout decision