[Journey] $0 Infra (Cloudflare Edge) + Claude Code + Programmatic SEO → Local SERP Domination in Colombia (Launch Phase)

[ninobareque]

Hi BHW,

I’m a Staff software engineer (10+ years XP), and instead of building another bloated Shopify store, I’ve spent the last few weeks building a programmatic + edge-first SEO system to dominate a high-ticket niche in a LATAM market (Colombia) — SUS-304 kitchen hardware.

This is not churn & burn.
I have real imported inventory, local fulfillment, and I’m building for scale.

I’m approaching this with a simple belief:

→ Engineering + speed of execution can outperform traditional SEO playbooks.

Curious to see if that actually holds.

---

The Stack ($0 Infra, Fully Live)​

• Edge Infra: Cloudflare Workers + Pages
• Core System: Built using Claude Code (agentic workflows for generation + structure)
• Analytics: PostHog (deep funnel + behavioral tracking)
• Architecture: No CMS, no plugins, no bloat

Performance:

• TTFB < 100ms
• LCP ~1.2s
• CLS = 0
• Infra cost: $0/month

---

Why $0 Infra Actually Matters (vs Shopify / Typical Stacks)​

This is not just about saving money — it’s about removing structural limitations.

Typical setup with platforms like Shopify:

• Monthly subscription ($39–$399)
• Multiple paid apps (reviews, SEO, CRO, speed) → $50–$300+/month
• Increasing JS bloat from plugins
• Slower performance as the stack grows
• Limited control over core architecture

Even “serverless” alternatives often introduce:

• Hidden costs (databases, functions, bandwidth)
• Performance inconsistencies
• Fragmented systems that don’t scale cleanly

---

With this edge-first system:

• Cost ≈ 0 regardless of traffic → I can scale without financial pressure
• No plugin overhead → performance stays consistent as I grow
• Full control of HTML output → precise SEO execution
• No platform constraints → I can build exactly what the use-case needs
• Instant deployment at the edge → faster iteration cycles

---

The key difference:

→ In most setups, scaling traffic increases cost and complexity.
→ In this setup, scaling traffic only increases opportunity.

---

This changes how I approach SEO entirely:

• I can test aggressively
• Launch hundreds of pages without risk
• Iterate faster than competitors

Most importantly:

→ I’m not optimizing within constraints — I’m defining them.

---

My hypothesis:

Most Shopify-based stores are structurally limited before they even start competing.

Curious if anyone here has seen the same — or thinks this doesn’t actually matter in rankings.

---

Engineering System (This is the real play)​

Everything is standardized and generated through controlled inputs:

• SEO checks: semantic structure, indexability, internal linking
• CRO checks: CTA placement, above-the-fold clarity, friction audit
• PSEO variables: location, modifiers, intent clusters
• Reusable templates: generated via structured logic (not blind page generation)

Claude Code is used to:

• Generate page structures
• Enforce consistency across templates
• Scale variations without introducing technical debt

One thing I’ve noticed already:

→ Without strict standardization, PSEO becomes unmanageable very fast.

Trying to keep this high-level for now and share deeper details as things evolve.

---

Strategy (Phase 1 → Local Capture)​

• Fresh domain (.com)
• Programmatic landing pages:
  
  • Product × City
  • Category × City
  • Intent variations (not just keyword swaps)

I’m starting with one smaller city (not disclosed) where SEO competition is still low and most players are not technically optimized.

Goal:

→ Capture local demand first
→ Build strong positioning in a controlled environment
→ Expand into top 10 cities after validation

• Planning to launch Google Ads after initial indexation (~1 month)

---

Current Status (Launch Phase)​

• Site: LIVE
• Build time: ~few weeks
• Indexed pages: starting now
• Traffic: planning to start paid acquisition in ~1 month after initial indexation
• Reviews: will be acquired gradually (real customers only)

(I’ll share proof and deeper metrics in upcoming updates)

---

Questions for those scaling aggressively​

1. Sandbox / Trust Acceleration:
   How much impact have you seen from combining strong UX + immediate paid traffic on fresh domains?
2. Authority in LATAM:
   Fastest clean ways to build authority links in markets like Colombia?
3. Backlinks Strategy:
   Is it still worth buying backlinks in 2026 for local niches, or are you seeing better ROI with PR / real mentions?
4. Analytics Tradeoff:
   Using PostHog early vs going fully server-side for performance?
5. PSEO Scaling Reality (2026):
   Where are you seeing diminishing returns — volume vs variation?
6. Local Domination:
   Highest ROI actions to win a single city fast?

---

End Goal​

• Dominate 1 city → expand to 10 major cities
• Keep infra near $0 while scaling
• Build a defensible SEO + conversion system

---

I’ll update this thread with:

• Indexation data
• Ranking movement
• Conversion metrics
• What actually works vs what breaks

If there’s interest, I can also break down:

• Claude Code workflows
• Edge architecture decisions
• PSEO generation logic

Also happy to answer any questions or go deeper into specific parts of the build.

Let’s see if engineering + execution can outperform traditional SEO — or if I’m completely wrong.

 

[Proxy Cake]

Wish you wouldn't have used AI to write this post but it's interesting nonetheless

 

[ninobareque]

Wish you wouldn't have used AI to write this post but it's interesting nonetheless

Fair point

English is not my first language so I tend to over-structure things to be clear.

Will keep future updates more raw and focused on real data.

I have high expectations for this project — let’s see how far it can go.

Appreciate the feedback

 

[Topiano]

Looks like a structured goal . Are you by chance working with expired or on a fresh domain ?

 

[ninobareque]

Looks like a structured goal . Are you by chance working with expired or on a fresh domain ?

It's a fresh domain. I actually didn't go for an expired one this time; I wanted to see how much weight a clean EMD (Exact Match Domain) still carries when paired with a solid technical setup. I’m currently testing a serverless architecture (Cloudflare Workers/Pages) to serve a multi-city pSEO strategy. My theory is that the high semantic relevance of the EMD, combined with top-tier performance and clean JSON-LD, might help bypass or at least shorten the typical sandbox period. I'm still tweaking the UI/UX, but the core engine is ready. What’s your take on going full EMD for local niches these days? Do you think the structured data and edge performance can compensate for the lack of initial domain authority?

 

[Topiano]

It's a fresh domain. I actually didn't go for an expired one this time; I wanted to see how much weight a clean EMD (Exact Match Domain) still carries when paired with a solid technical setup. I’m currently testing a serverless architecture (Cloudflare Workers/Pages) to serve a multi-city pSEO strategy. My theory is that the high semantic relevance of the EMD, combined with top-tier performance and clean JSON-LD, might help bypass or at least shorten the typical sandbox period. I'm still tweaking the UI/UX, but the core engine is ready. What’s your take on going full EMD for local niches these days? Do you think the structured data and edge performance can compensate for the lack of initial domain authority?

Had decent success working with EMD local niche and still do till date in age of AI.

I’d always advice not to burn your index debt so early and restrict the amount of pages you plan creating / indexing on daily otherwise this would be tagged total spam. I’d strongly advise getting featured/ mentioned on local news . That’s your PR success especially if targeted to your geo location.

 

[ninobareque]

Had decent success working with EMD local niche and still do till date in age of AI.

I’d always advice not to burn your index debt so early and restrict the amount of pages you plan creating / indexing on daily otherwise this would be tagged total spam. I’d strongly advise getting featured/ mentioned on local news . That’s your PR success especially if targeted to your geo location.

Thanks for the heads-up! Your timing is perfect as I’m just about to finalize the deployment logic.

I completely agree on the 'index debt' risk; it’s easy to get carried away when the tech allows for such scale. From a dev perspective, I’m building automated scheduling into my pipeline to ensure a steady, natural-looking rollout.

The plan is to use pSEO for long-tail keywords with strict rewriting rules to keep the quality high and the footprint clean. Using this approach,

I can generate a reasonable amount of pages per product with just a few clicks, but I'll be taking your advice and throttling the output to stay under the radar while focusing on that local PR for entity validation.

In a few days, I’m planning to upload my CI/CD architecture diagram. I think it will give you a better perspective on how I’m handling the scalability of the project without compromising the domain’s health. Truly appreciate the insight!

 

[CTRPhantom]

EMD can help a bit if you actually back it with real signals, but it’s not magic. Fresh domain + edge is cool, but watch the index debt—don’t dump pages faster than you can quality-control them. Curious how Claude Code handles multi-city templates in practice. Keep us posted with results, this is an interesting experiment.

 

[ninobareque]

EMD can help a bit if you actually back it with real signals, but it’s not magic. Fresh domain + edge is cool, but watch the index debt—don’t dump pages faster than you can quality-control them. Curious how Claude Code handles multi-city templates in practice. Keep us posted with results, this is an interesting experiment.

Thanks for sharing your thoughts! You nailed it—the whole point of this experiment is to see how a fresh domain reacts to a high-performance Edge setup from scratch. Regarding Claude Code, it’s not just about 'AI generation' for me; it’s about software architecture. I’ve built standardized, globalized scaffoldings where, based on CRO rules, the system generates multiple versions of the HTML content, randomizing positions and interpolating new text strings into a pre-established base.


Since these are product pages, the goal is to rotate the entry hooks and re-structure descriptions using pSEO long-tail logic. I’m still refining the interpolation logic—I’ve even considered adding another LLM layer for a full-page rewrite to keep things fresh—but I truly value your point about quality control. My ambition for speed shouldn't come at the cost of being flagged as spam.


Currently, the site is live at $0 cost with a fully automated pipeline. In my next update, I’ll share the architecture diagram so you can see the 'how' behind the deployment. Hopefully, in a few weeks, natural indexing will start doing its job. Stay tuned!

 

[ninobareque]

UPDATE DIA 1 AFTER FIRST DEPLOY

technical Update: Workflow & Architecture for the EdgeForge pSEO Engine​

"Technically, the architecture is solid. Mobile performance scores are optimal; the focus now shifts to conversion and sales (CRO)."

1. DEVELOPMENT & DATA SOURCE (THE BLUEPRINT)​

• Source Control (Git): The starting point is an ultra-lightweight codebase using Vanilla HTML5 and Tailwind CSS, ensuring maximum loading speeds and a clean DOM.
• Data Source: A centralized JSON file acts as the "Single Source of Truth" for all data powering the pSEO (products, pricing, variants, etc.).
• AI Code Integration: An AI assistant aids in generating and validating content rewriting rules, ensuring every page variant is unique, high-quality, and avoids footprints.

2. CI/CD PIPELINE (THE ASSEMBLY LINE)​

• GitHub Actions: Full Integration and Continuous Deployment. Every git push triggers automated tasks.
• Technical Audit: Before deployment, the system validates JSON integrity and checks the sitemap's validity. If a rule fails, the build stops.
• Deploy to Edge: Once validated, code and data are instantaneously distributed across the global edge network.

3. SERVERLESS DEPLOYMENT & DELIVERY (THE ENGINE)​

• Cloudflare Workers/Pages: We avoid traditional hosting. User requests are processed instantly by a Worker at the edge.
• Worker Logic: The Worker handles routing, queries the JSON data, and assembles the final static HTML in milliseconds.
• Throttling & Sitemap Control: Following the forum's advice, I’ve implemented a Staged Sitemap Rollout. This logic releases pages gradually to manage index debt and prevent spam flags from Googlebot.
• Global Edge Network: Ultra-fast delivery worldwide, with a specific geographic focus on Colombia.

4. INDEXING & VALIDATION (SIGNALS)​

• Screaming Frog Audit: Automated audits verify that generated pages meet Status Code 200 requirements, unique metadata, and valid Schema.org markup.
• SEO Signals: Technical performance is backed by a current mobile score of 92/100 (Performance), 100/100 (Accessibility), and 92/100 (SEO).
• Next Steps: While a few mobile UI tweaks are still pending, the infrastructure is ready. External validation will now be complemented by tracking indexing and pursuing Local PR mentions to strengthen domain authority.

 

[ninobareque]

UPDATE DAY 2 .

Current Milestone:
After deep-diving into performance optimization, I’ve finally solved all mobile rendering issues. Our Core Web Vitals are now in the green, and the impact was almost immediate: Google has already indexed the first pages of the initial structure. The SEO foundation is officially solid.

️ The Next Move: Scalable Dynamic Workers
Now that the baseline is stable, it's time to scale. Instead of bloating the server with thousands of static HTML files, I’ve decided to move toward a Live Edge-Rendering architecture using Cloudflare Workers.

The Strategy: A dedicated Frontend Worker will act as a real-time rendering engine.
The Workflow: No physical .html files will exist for city-specific pages. The Worker will intercept requests (e.g., /location/product-slug), pull local data from a JSON/KV store, and "build" the page on the fly with localized content, neighborhoods, and regional terms.
Security & Decoupling: I am keeping my Checkout Worker completely separate to ensure payment stability while the new pSEO Engine handles the traffic growth.


Going "Deep Work" for a Month
I’ll be going quiet for about a month to focus on building this engine and allowing Google’s algorithm to fully crawl and stabilize the current routes. I’ll return with data on how the traffic behaves once we launch the first 160+ dynamic city pages.


Thanks for following the journey! Time to turn this faucet on full blast.

 

[ninobareque]

UPDATE 3

I’m excited to share some major technical milestones in my current journey. We’ve moved past the architecture phase and are now into the "Continuous Deployment" of our content engine.

Technical Milestones​

I’ve officially finished the deployment of two core systems running on Cloudflare Workers:

• pSEO Engine: Fully operational and optimized for zero-latency routing.
• Article Worker: Successfully implemented a content logic that maintains a high density of information, effectively eliminating "Thin Content" issues.

Early Performance & Indexing​

The data is starting to roll in, and the initial logs are looking very promising:

• Indexing: We already have 20 pages indexed by Google.
• Rankings: Most of these new pages are already appearing in Position #5 or higher right out of the gate.
• GSC Cleanup: All legacy conflicts and errors in Google Search Console have been resolved. The architecture is now 100% "Zero-Legacy."

Local Authority & Expansion​

• Social Proof: We’ve hit 10 verified reviews on Google Business Profile (GBP), strengthening our local trust signals.
• Next Step: I’m currently activating the Google Shopping tab to integrate our inventory directly into the search ecosystem.

The 90-Day Roadmap​

Consistency is key. For the next 3 months, I will be reporting on:

• Publishing Cadence: Deploying 3 to 4 new high-quality URLs per week.
• Metric Monitoring: Tracking the correlation between new indexations, impressions, and CTR.

Building this has been a challenge, but seeing the first pages hit the top of the SERPs validates the entire stack. Let's keep building!

18/04/2026




23/04/2026



fix problems



pages



27/04/2026



I am fully aware that the site is still in its 'stabilization phase.' It hasn't even been a month since the initial launch, so I am monitoring the logs very closely.For now, the operation remains incredibly lean—hosting and infrastructure costs are virtually zero thanks to Cloudflare, with the only real expense being the $30 USD/year for the domain. My current focus is winning the local game first. SEO in my home city isn't as competitive or 'dedicated' as in the major capitals, so I’m establishing a solid base here before scaling our pSEO routes to the rest of the country. One step at a time.

 

[Larry Igna]

The Worker will intercept requests (e.g., /location/product-slug), pull local data from a JSON/KV store, and "build" the page on the fly with localized content, neighborhoods, and regional terms.

I don't know how much data you (will) have, but D1 database is free (requirement met) and it allows for more flexibility if system grows. KV store is good enough for simpler stuff, but with D1 you get a powerful SQL DB.
Otherwise, solid plan, wishing you good luck!

 

[rampzeus]

This is an interesting journey, I am also doing large scale local seo. But mine is many sites on a single vps and also affiliate based not own product.

Will you scale this to multiple domains or bet all on one domain?

How will you handle the payments and PII?

With my experiments with products most people who search product city seek a physical store not online. Might be different in latam market. Keep it up.