The technical layer most agencies outsource.
Most SEO reports stop at "improve page speed" and "fix crawl errors." That's where this starts. Cloudflare Workers and cache rules, Core Web Vitals diagnosed against field data instead of a lab score, redirect and crawl architecture, structured data — found and fixed by the same person, because I write the code.
Sound familiar?
- Your audit says "reduce render-blocking resources" and nobody can tell you which resource, or who owns it
- The site got faster in Lighthouse and nothing moved in Search Console
- Some visitors see a broken page and others see a fine one, and no one can reproduce it
- Your redirects "work" — through four hops, losing a little authority at each one
- You're paying for a CDN and nobody has opened the cache rules since it was switched on
- Your developer says it's an SEO problem, your SEO says it's a developer problem, and you're paying both
What the work covers
Edge & CDN configuration
Cloudflare Workers, Snippets, Transform Rules, Cache Rules, and WAF — the layer between your server and your visitor. Cache behaviour that matches how your site actually deploys, purge automation so a release can't serve stale assets, and header and redirect logic handled at the edge instead of bolted into the application.
Core Web Vitals, diagnosed properly
Lab scores are a starting point, not the goal. The work traces LCP to the specific element and request that causes it, finds what's actually blocking the main thread, and eliminates layout shift at the source rather than chasing a number. Framed against field data where field data exists.
Crawl, indexation & redirect architecture
What Google can reach, what it chooses to index, and what it wastes budget on. Redirect chains collapsed to single hops, canonical and hreflang logic that holds up at scale, robots and sitemap directives that agree with each other, and legacy URL patterns retired without dropping the equity they carry.
Structured data & AI-search readiness
Schema built as a connected entity graph rather than disconnected blocks — augmented by @id, never redefined per page. Plus the things AI search actually reads: clean server-rendered HTML, an accurate llms.txt, and crawler access verified by request logs rather than assumed.
Three things that actually happened
Not case-study gloss. These are documented problems with documented fixes, including the parts that don't flatter me.
A CDN edge serving a stale file under a brand-new URL
A site update shipped and some visitors saw an unstyled page while others saw a perfect one. The stylesheet was cache-busted with a version string, so by every normal rule this was impossible — a new URL cannot be a stale cache hit.
It was. At least one Cloudflare PoP had cached an old stylesheet body under the new versioned URL, a race between the HTML and the asset going live during deploy. A 30-day max-age then pinned that bad copy for a month for every visitor routed through that edge. Post-deploy verification passed because it happened to hit a different, clean PoP — which is why it survived a check that should have caught it.
Diagnosis was a two-line comparison: fetch the versioned URL and a randomised one, compare cf-cache-status and payload. Versioned URL a HIT with stale bytes, random URL a MISS with correct bytes — the edge is poisoned, not the build. The fix was a never-reused timestamped version string, the asset TTL capped at one day with stale-while-revalidate, and a purge-on-deploy action so the edge can't outlive a release.
Why this one is here: it's a failure mode that looks like a code bug, survives ordinary verification, and is invisible to every SEO tool on the market. Diagnosing it requires knowing what a CDN does — not what a crawler reports.
A motion system that adds nothing to CLS
Animation is usually where layout stability goes to die. This site's motion system was built so nothing animates into reserved space: count-up figures pre-measure their final width, the audit panel ships its finished state in the HTML, and reveals are restricted to opacity and transform. The whole redesign added roughly 1.5 KB of JavaScript.
Read these honestly: mobile Lighthouse medians across three page types, multiple passes with the worst-LCP outlier dropped. They are lab measurements of a current state — not a before/after, and not field data. Lab numbers are a proxy; CrUX field data is what Page Experience actually uses. I'd rather show you a measured number with its limits attached than a percentage improvement I can't substantiate.
188 legacy URLs retired without losing the equity
A Drupal 9 site on Pantheon carrying years of accumulated URL patterns, competing language variants, and database-level URL regressions that reappeared after each fix. The work retired the legacy patterns, shipped hreflang that survived the CMS regenerating its own aliases, and resolved the regressions at the layer that caused them instead of patching the output.
How it works
Start with the health check
$500 flat. The cheapest way to establish whether your problem is technical at all.
Diagnosis, with evidence
Findings tied to specific requests, rules, and headers — not a severity score from a crawler.
Scope the fix
Tiered by remediation effort so you can greenlight by tier instead of all-or-nothing.
Implementation
I ship it, verify against production, and document what changed. The $500 credits toward work booked within 60 days.
When I'm not the right fit
If your site is technically sound and your problem is that nobody links to you or your content is thinner than the competition's, technical SEO will not fix that. I'll tell you after the health check rather than sell you infrastructure work you don't need — a clean bill of technical health is a legitimate outcome, and you keep the report either way.
If you need enterprise-scale log-file analysis across millions of URLs, or a full-time in-house technical SEO, I'm the wrong shape of help. This works best for sites in the range where one person can hold the whole system in their head and act on it quickly.
And if you want content strategy, link building, or paid media, that's not what this is. I do the technical layer and I'd rather be excellent at one thing than adequate across five.
Working in Kentucky specifically? The Kentucky SEO services page covers the same work with local market context. If you'd rather start by understanding what an audit finds, the $500 site health check is the entry point for everything above.
Frequently asked questions
What counts as technical SEO?
Everything that decides whether a search engine can reach, render, and understand your pages — and how fast a real person gets to see them. Crawling and indexation, redirects and canonicals, site architecture, Core Web Vitals, CDN and cache configuration, structured data, and the analytics that prove any of it moved. It's the layer underneath content and links, and when it's broken, content and links underperform for reasons no keyword tool will show you.
Do you work on Cloudflare directly, or just make recommendations?
Directly. Workers, Snippets, Transform Rules, Cache Rules, WAF, and purge automation are part of the deliverable, not a list of suggestions handed to someone else. That's the difference between this and most technical SEO engagements — the person diagnosing the edge is the person who writes the Worker. I've built production Workers with signed webhooks and KV state, not just cache rules.
How is this different from an SEO audit?
An audit tells you what's wrong. This fixes it. Most engagements start with the $500 site health check because it's the cheapest way to find out whether your problem is technical at all — plenty of sites don't need this service and I'd rather say so than sell it. If the health check surfaces infrastructure problems, the fix work starts from there and the $500 credits toward it within 60 days. You can book the health check here.
My site scores well in PageSpeed Insights. Do I still need this?
Maybe not — and a lab score of 95 is a genuinely good sign. But lab scores are simulated on one device on one connection, and Google's Page Experience signals use CrUX field data from real visitors. Sites regularly pass in the lab and fail in the field, usually because of what happens on repeat visits, on real mobile hardware, or at a specific CDN edge. If your lab score and your field data disagree, that gap is exactly the work.
Can you guarantee rankings?
No — and technical SEO is the area where that promise is most obviously false. Fixing infrastructure removes the reasons a page can't rank. It doesn't manufacture demand, authority, or better content than the competition. What I can promise is that the diagnosis is real, the fix is implemented rather than recommended, and you get a written record of what changed and why.
Find out if the problem is underneath the content.
$500 tells you what's actually wrong at the technical layer, in plain English, with the fixes tiered by effort so you can decide what's worth doing. The report is yours whether we work together after that or not.
Start the technical audit.
Tell me about the site, the stack, and what's been frustrating you. I'll confirm fit and timing within one business day — and if the problem isn't technical, I'll say so before you spend anything.