Results
What I have actually moved, and what I am moving in public right now
Two products with real revenue numbers, and one live experiment you can check yourself: this site, which spent five months invisible to every AI crawler and is now being rebuilt in the open.
What are the numbers?
Eight years in WordPress and SaaS marketing, most of it on the revenue side of commercial plugins. Two products I worked on, with the before and after I can walk you through line by line on a call:
| Product | What changed | MRR before | MRR after |
|---|---|---|---|
| CTX Feed | Product-led content, comparison pages, marketplace positioning | $4,000 | $23,000 |
| Saasland | Organic growth programme — traffic up 200%+ over the same window | $300 | $19,000 |
Why are there only two?
Because those are the two where I owned the number. I could pad this page with logos of companies whose sites I touched for a month, and every agency page you have ever read does exactly that. Two results you can interrogate are worth more than twenty you cannot.
Both are pre-GEO work — content, comparison pages, positioning. The mechanism is the same one that decides AI citations; the engines were just not the distribution channel yet.
What is happening on this site?
The most honest case study I have is the one I am running on myself, in public, starting from a genuinely bad baseline.
In September 2026 I audited softrecs.com with the same thirty checks I sell. The finding that mattered was not a content finding. The TLS certificate had lapsed on 5 April 2026 and stayed lapsed for 152 days. Every AI retrieval crawler — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot — fails the handshake before it reads a single byte of HTML. The site was not ranked badly. It was unreadable.
That is the starting line, published before the result is known. The certificate is reissued, the site is being rebuilt around retrievability, and the prompt set is being run monthly from here. If it works, this table will show it. If it does not, this table will show that too — and a GEO consultant who quietly deletes his own baseline is not one you should hire.
Nothing yet. It is a baseline, not a result. The point of publishing it early is that you can check the arithmetic later instead of taking my word for it.
Why does a lapsed certificate matter so much?
Because retrievability sits underneath everything else, and almost nobody checks it. Schema, headings, an llms.txt file — none of it is read if the connection is refused. Ahrefs looked at 137,000 sites and found 97% of published llms.txt files had never been fetched by anything at all. The one controlled test of schema markup found no measurable lift in AI Overview citations.
Meanwhile the things that do correlate are unglamorous: brand mentions at 0.66 against backlinks at 0.19, freshness, original data with sources, and being served as HTML that a non-JavaScript crawler can read. That is the order I work in, and this site is the demonstration.
When will there be client case studies?
When there are results worth publishing and a client who agrees to be named. Audits take two weeks, the changes take a month to ship, and the engines take another month or two to re-crawl and re-rank. Anyone promising you a case study faster than that is showing you a screenshot, not a result.
Want your own baseline?
Send me your wordpress.org slug and I will run the five prompts your buyers use, and send back what the engines actually said. Free, no call required, usually within a day.