SEO Experiments I've Run: What Actually Works and What Doesn't

I'm Venetsian Jakimov, and I've been poking, prodding, and occasionally breaking websites for 20+ years from Bulgaria. Here's a modern retrospective on the experiments that shaped my thinking — what worked, what flopped, and what I'd test again in 2026 without flinching.

The Subdomain Caper: A Blast From the PageRank Past

Many years ago, I spun up blogs and directories on subdomains of a primary site, then fed them into roughly a thousand web directories. The goal was to watch how Google treated the links from those subdomains: internal boost or external juice?

The takeaway: Google largely treated those subdomain links as internal. The hoped-for "external" authority bump didn't materialize, and most of those subdomain pages drifted into the supplemental index. Building a constellation of thin subdomains won't trick the graph. It taught me to stop chasing loopholes and start investing in content and architecture that earn links on merit.

Domain Authority Now vs. The PageRank Era

Back then, visible PageRank made SEOs obsess over that green bar. Today, Google still relies on links, but the story is broader and subtler: page-level importance, site-level trust signals, topical depth, user satisfaction, and a ruthless willingness to ignore low-value stuff. Think in systems: your internal links, content quality, technical health, and reputation form a compound signal. You don't "win" with a single hack — you accumulate advantages across the board.

Subdomain vs. Subdirectory in 2026

Both can work. In practice, subdirectories are usually simpler for consolidating signals, sharing crawl equity, and keeping analytics tidy. If your blog, docs, or resources are part of the same brand experience, a subdirectory often makes integration easier. Subdomains still make sense for truly distinct apps, regions with separate operations, or technical reasons. If you use subdomains, treat each as its own property in Search Console, interlink thoughtfully, and avoid spinning up thin microsites.

Modern Experiments Worth Running

A/B Testing Titles and Descriptions

CTR is a lever. Create two well-formed title variants that reflect the same content. Run them in matched groups: similar pages, similar intent. Measure impressions, CTR, and average position via Search Console; monitor conversions in analytics. Keep tests at least two weeks and change only one element at a time.

Schema and Rich Result Impact

Add or refine structured data (FAQ, HowTo, Product, Article). Your KPI isn't rankings; it's SERP real estate and CTR. Track eligibility in Search Console's Enhancements reports and compare CTR before/after. Watch for cannibalization: an FAQ on the SERP that answers everything can slash pageviews. Sometimes fewer clicks but higher qualified traffic is the real win.

Internal Linking Restructuring

Internal links are the quiet powerhouse. Map your current click depth and identify orphaned or buried pages. Build topic hubs: cornerstone guides linking to focused subtopics, with subtopics linking back up and across. Add contextual links where users need them. Measure changes in crawl stats, average position of target clusters, and time to index.

Content Pruning and Consolidation

Audit for thin, overlapping, or obsolete content. Merge duplicates into a single stronger URL, 301 the rest, and update internal links. For truly dead pages with no links or traffic, remove and return 410/404. The goal isn't fewer URLs; it's a higher density of value. Track total impressions, average position across the pruned cluster, and crawl efficiency.

The Real ROI: An Experimental Mindset

Tools change; principles don't. Form a hypothesis, define success upfront, isolate variables, and give tests enough runway. Document everything. Celebrate being wrong quickly — it means you learned cheaply. Share results with your team so decisions compound.

Two decades in, I don't chase loopholes; I chase clarity. Build for users, make it easy for crawlers, prove your expertise, and iterate with curiosity. The rankings follow the work.