I’ve been blogging since the 2000s. I’ve watched SEO evolve from keyword stuffing to algorithmic complexity. I’ve seen trends come and go. And after decades of trial and error, I’ve learned one thing:
There is no universal system that works for everyone.
What I’m about to share isn’t a blueprint for you to copy. It’s how I built MY system for MY five blogs through experiments, failures, and small wins that eventually compounded into steady growth.
Your system will look different. Your bottlenecks will be different. Your solutions will be different.
But the PROCESS of building a system? That’s repeatable.
The Five Blogs (Same System, Different Niches)
I run five blogs:
- QAJourney.net Quality assurance and testing methodologies
- EngineeredAI.net AI tools and techniques
- RemoteWorkHaven.net Remote work strategies
- MomentumPath.net Productivity and mindset
- HealthyForge.com Health and wellness
Same WordPress setup. Same hosting. Same content creation workflow. Same SEO optimization process.
The only difference? Topic focus.
And here’s the thing: they all had different problems. Google loved some, buried others. Bing deindexed one completely. AdSense approved HealthyForge first, then EngineeredAI. But RemoteWorkHaven, MomentumPath, and QAJourney? Rejected multiple times for “low-value content.”
Same quality. Same editorial process. Different outcomes.
So I had to build a system that worked regardless of what search engines thought of my domain names.
The Experiments: What I Tried Before Finding What Worked
I didn’t start with a system. I started with questions and experiments.
On content creation:
- How do I make AI-generated drafts sound human? Built an editorial auditor prompt that removes AI fingerprints.
- How do I get better outputs from AI? Learned to ask better questions with clarity, context, and constraints.
- How do I make AI challenge my ideas instead of just agreeing? Engineered friction into prompts to get real feedback.
- What’s my actual content workflow? AI drafts, I curate, AI refines, I polish.
- How do I create visuals without design skills? ChatGPT generates prompts, Gemini creates images.
- Which image compression tools actually work? Tested everything, use Upscayl for upscaling and Squoosh for compression.
On SEO and discovery:
- How do I get discovered by LLMs, not just Google? Schema markup, clean structure, canonical links.
- Can I target specific geographic regions with LLM optimization? Yes, with proper structured data.
- Why is Google burying my AI blog but not my others? Domain name discrimination is real and documented.
Some experiments worked. Some didn’t. Some worked but only in specific situations.
The key was tracking what stuck.
What Stuck: The System That Emerged
After months of blind writing (December 2024 to August 2025), failed AdSense applications, Blogorama scraping my content and getting indexed first, and watching EngineeredAI thrive on Google but get completely delisted by Bing (while my other four blogs ranked normally on both), I finally had enough data to see patterns. Interestingly, DuckDuckGo started showing my content, and AI bot visibility increased significantly. Even more interesting, Perplexity and Copilot know EngineeredAI exists. If you do an AI search on Bing through Copilot, it sees the site even though regular Bing search delisted it.
Here’s what compounded into my current system:
Content Creation Layer
- AI generates the first draft (I use ChatGPT, but now testing GPT4All locally to save tokens)
- I curate by stripping out fluff, fixing structure, adding my voice
- AI refines through a pass with Claude or ChatGPT for grammar, clarity, removing AI fingerprints
- I polish with a final editorial pass, adding examples, ensuring it sounds human
- Visuals come from ChatGPT generating image prompts, then Gemini creates visuals, then Upscayl for upscaling, then Squoosh for compression
- Publish with proper schema markup and metadata
SEO/Discovery Layer
Schema markup with LLM-friendly structured data, geographic targeting where relevant
Meta keywords (controversial, but hear me out) that serve NOT for ranking, but as categorical signals with homepage broad topical keywords and per-post semantic keywords for specific content categorization (keeps my categories and tags taxonomy clean and rich, avoids keyword cannibalization since Google often indexes category/tag pages before individual posts)
- Homepage optimization through strategic structure, clear navigation, ad placement above the fold
- Data infrastructure using GA4, Google Search Console, Cloudflare Analytics to track everything
- Internal linking that connects related content across posts
- Multiple discovery systems because you can’t rely only on Google (LLMs, syndication to Dev.to/Hashnode, GitHub Gists with canonical links)
Why Meta Keywords?
Both Claude and ChatGPT told me meta keywords were useless. “Google deprecated them.” “They don’t help ranking.”
They were right about ranking. But wrong about uselessness.
Google deprecated meta keywords as a RANKING FACTOR because people abused them. But Google still PROCESSES them. Search engines still USE them for categorical understanding and search appearance.
So I use them as organizational markers with homepage meta keywords for broad topical signals and per-post semantic keywords for specific content categorization (keeps my categories and tags taxonomy clean and rich, avoids keyword cannibalization since Google often indexes category/tag pages before individual posts).
It’s not for ranking. It’s for better search appearance and helping search engines understand site structure.
For the three rejected blogs (RemoteWorkHaven, MomentumPath, QAJourney), I had to add syndication to Dev.to/Hashnode for external validation and category/tag cleanup by removing redundant/confusing taxonomies.
After that? AdSense approvals started coming through. QAJourney was the last to get approved, but is now the one earning the most in AdSense from $0.05 in two months to $0.02-0.08 per day.
The meta keywords on homepage plus semantic keywords on posts came much later (early October 2025). That’s when the traffic growth really kicked in with 66.7% new visitor increase after implementation.
Two different problems, two different solutions:
- AdSense approval equals syndication plus taxonomy cleanup
- Traffic growth equals meta keywords plus semantic keywords for better search appearance
That’s not coincidence. That’s giving algorithms clearer signals about what your content actually covers.
The Results: Small Wins That Compound
Here’s what happened after implementing this system (late September through early October 2025):
Traffic Growth:
New visitors up 66.7% in one month, total impressions jumping from 193 to 996, and total clicks rising from 2 to 12.
Engagement Quality:
Top performing post showing 55 pageviews, 41 sessions, 58.54% engagement rate, 1m 29s average session duration. Multiple posts hitting 50% plus engagement rates. Content about mental structure, productivity, and systems performing best. Average CTR holding at 1.2% while impressions increased.
Search Visibility:
Posts ranking for target keywords despite initial suppression, geographic targeting working, and average position improving from “discovered but not indexed” to actual ranking.
These aren’t huge numbers. But they’re steady. Repeatable. And they compound.
How to Build YOUR System (Not Mine)
Here’s the process I used. This part is repeatable, even if your solutions will be different:
Step 1: Write First, Optimize Second
Don’t start with SEO. Start with content that matters to your niche.
I wrote blind for months. No data. No rankings. Just relevant content about topics I understood.
That foundation matters. You can’t optimize nothing.
Step 2: Set Up Data Infrastructure
You can’t build a system without measurement. Set up Google Search Console for impressions, clicks, CTR, and average position. Add GA4 for engagement, session duration, and bounce rate. Use Cloudflare Analytics for bot/crawler behavior. Include keyword research tools for search volume and competition.
Wait 2 to 4 weeks. Watch the patterns emerge.
Step 3: Identify YOUR Bottleneck
My bottleneck was domain name discrimination plus Blogorama scraping.
Your bottleneck might be not getting indexed at all, getting indexed but not ranking, ranking but no clicks, clicks but no engagement, or something completely different.
The data will show you where you’re blocked.
Step 4: Test Solutions for YOUR Situation
Once you know your bottleneck, test solutions. Not getting indexed? Try schema markup, sitemap optimization, manual index requests. Not ranking? Try better keyword targeting, content depth, internal linking. No clicks? Try better titles, meta descriptions, featured snippets optimization. No engagement? Try better content structure, clearer writing, relevant examples.
I tested meta keywords because my bottleneck was categorical understanding. You might need something completely different.
Step 5: Keep What Works, Drop What Doesn’t
Track everything. If something moves the needle, keep doing it. If it doesn’t, drop it.
For me, schema markup worked, meta keywords as categorical signals worked, homepage optimization worked, and AI drafts plus human curation worked. Writing high-intent viral topics from social media trends didn’t work (no engagement, just noise). Over-optimizing for search volume without audience fit didn’t work. Expecting Google to rank AI content without engineering discoverability didn’t work.
Step 6: Repeat the Process Across Projects
Once you’ve built a system that works for one project, apply the same PROCESS to another:
Write relevant content, measure performance, identify bottlenecks, test solutions, keep what works, then iterate.
The system is the process, not the tactics.
Why Small Wins Matter More Than Big Wins
People want the one trick that 10x’s traffic overnight.
That’s not how systems work.
Systems compound small wins over time. A 66.7% new visitor growth sounds small, but month over month, that’s significant. 1m 29s engagement on one post sounds small, but that’s real readers, not bots. 12 clicks instead of 2 sounds small, but that’s a 500% increase.
The compound effect is what matters. Not the individual number.
What I’m Still Learning
This system isn’t finished. I’m still experimenting with testing GPT4All locally to reduce token costs, exploring whether syndication to Dev.to/Hashnode actually helps discovery, watching which content themes perform best across different blogs, figuring out which schema markup types matter most for LLM discovery, and testing whether multiple discovery systems (Google, Bing, LLMs, social) compound or cannibalize.
Every project teaches me something new about what works and what doesn’t.
But the system itself, measure, identify bottlenecks, test solutions, iterate, that’s solid.
The Honest Reality
I’m not an SEO expert. I’m someone who’s been blogging since the 2000s and figured out what works through trial and error.
What works for me won’t work for everyone. But the PROCESS of building a system works for anyone willing to measure, test, and iterate.
If you’re struggling with SEO right now, you probably don’t have a content problem. You have a visibility problem.
Start by identifying YOUR bottleneck. Then engineer YOUR solution.
Don’t copy my system. Build yours.
The small wins will compound.
Related experiments and explorations:
- LLM Optimization Guide
- Editorial Auditor Prompt
- Prompt Engineering for Critical Thinking
- Asking Better Questions
- How I Use Generative AI for Content
- Search Engine Discrimination Against AI Content (With Data)
- Geographic LLM Targeting Guide
Runs five blogs with the same system, different results.
This article documents real experiments across QAJourney, EngineeredAI, RemoteWorkHaven, MomentumPath, and HealthyForge.
Tests what actually works in SEO and content creation, not what guru courses promise.
Breaks down meta keywords, schema markup, and LLM optimization through trial and error since the 2000s.
If something doesn’t move the needle in real data, it gets dropped.
