Live SEO Case Study: This Site, Measured in Public

Why I run my own site in public

Every consultant says their results are real. A hospital buyer can’t check that, because client names and dashboards sit under NDA. I asked ChatGPT and Perplexity why someone should not hire me, and the first answer was exactly that: the numbers can’t be verified from the outside.

So this page is the proof you can check. My own site is a healthcare SEO business competing for healthcare SEO searches, and I publish its numbers from the day I rebuilt it. You can compare the numbers here with the site itself, with Google, and with what AI tools say about me when you ask them.

Day 0: where the site started

These are the numbers on 27 September 2026, the day the rebuilt site went live. Nothing is rounded up.

The search data told its own story. The queries with the most impressions were for orthopedic SEO, which came from one old page, and a company name that has nothing to do with me. Healthcare buyers searching for a consultant were not finding the site at all.

What an AI audit said about me

Before fixing anything I asked ChatGPT, in a temporary chat with personalisation off, to act as a US hospital group and list every reason not to hire me. It was useful, and some of it stung. This is what it found and what I did.

None of these were hard to fix. All of them were costing trust before anyone booked a call. Most healthcare sites I audit have the same kind of problem: small, visible inconsistencies that a careful buyer, or an AI model reading on their behalf, notices first.

What I changed, and why

1. Speed. Page caching and a clean server setup took the server response from 1.5 to 20 seconds to around 0.4 seconds on a cached page. Slow pages lose patients before they read a word.
2. One version of every page. All traffic now goes to one secure address, and every page names itself as the original with a canonical tag.
3. A clean index. The sitemap went from 70 URLs to 51. Tag, category and author archives were thin pages competing with real ones, so they are now kept out of the index.
4. Structure. Separate pages for each sector and each service, each owning one search phrase, linked the way a buyer moves from problem to proof to call.
5. Machine-readable facts. One connected schema graph for the business, the person and every page, plus an llms.txt file that tells AI tools what each page is for.
6. Content with sources. Fourteen guides on the questions healthcare buyers actually ask, each regulatory claim linked to its primary source, and one piece of original research on 30 lab websites.
7. Discovery. Sitemap submitted to Google, the six most important pages sent for indexing, and all 51 pages sent to Bing through IndexNow.
8. Email you can trust. SPF, DKIM and a new DMARC record, so replies to enquiries reach an inbox rather than a spam folder.

How I will report it

Every update uses the same measures and the same sources, so the numbers can be compared. If something goes down, it stays on the page with the reason.

What I won’t do is promise rankings. A new site in a competitive niche takes months, and my site is a consultancy, not a clinic. The point of this page is to show you the method and the honesty of the reporting, which is exactly what you would get as a client.

What this means for your practice

The same audit works on any healthcare site: pull the real baseline, ask AI tools the hard questions a buyer or patient would ask, fix what they find, and report on fixed dates. If you want to see how your site looks through that lens, book a free strategy call and I will run the same checks before we speak.