The Renewal Is Finished
After three months of work, the new website went live on September 25, 2026.
I started writing a website on WordPress in August 2011. I built a separate site for Rolfing in December 2015, merged the two in July 2017, and launched the English version in July 2019. Fifteen years of articles had piled up since then.
This is the final post in Building a New Website, a series in which I have been writing about the renewal as it went along. At the end of the previous post, I said I would write about what changed once the site went live. This is that post. What changed compared with the old website? Why did I change it? What evidence did I base the changes on? These are the three questions I want to answer here.
The Evidence Behind the Decisions
In an age without clear answers, with generative AI developing at a furious pace, how should a website be built? I think it is important to move forward with an answer of my own, a working hypothesis. In that sense, what I value about generative AI is that it analyzes a website’s access data carefully, and when something is hard to understand, it explains it in concrete words.
I based my decisions on data from three tools.
Google Search Console
A free tool from Google. It shows how many times this site appeared when people searched on Google, how many times it was clicked, what words people searched with, and where the site ranked in the results.
Bing Webmaster Tools
The same kind of tool for Bing, Microsoft’s search engine.
Bing is the other major search engine alongside Google. In Japan, most people search with Google. Bing, on the other hand, is the search engine built into Microsoft Edge, the browser that comes with every Windows computer, and it is strong in searches made from computers.
Bing is also connected to generative AI. Microsoft’s own AI, Copilot, uses Bing’s search results for its answers. OpenAI, the company behind ChatGPT, has made Bing an official search partner, and when ChatGPT searches the web to answer a question, Bing’s results are one of its main sources. Whether a page appears in Bing affects whether generative AI will cite it.
Microsoft Clarity
A free tool from Microsoft. If Google Search Console and Bing Webmaster Tools show what happens in search results, before anyone arrives at the site, Clarity shows what happens after they arrive.
It shows which pages were read, how long people stayed on each page, and how far down they read. Its heatmaps show, in shades of color, where on a page people clicked most and where they stopped reading.
Generative AI systems visit web pages automatically to build their answers. Clarity also shows which pages of this site AI systems such as ChatGPT and Claude read and how many times, and how many people came to the site after seeing an AI’s answer.
What the Data Showed
I have looked at all three every month since May. Three things became clear.
First, this site did not come up near the top when people searched for “ロルフィング” (Rolfing) in Japanese. As of May 2026, it ranked 41st in Google’s results.
Second, nearly everyone who comes to this site from search comes from Google, about seven times as many as from Bing. My other site, Mind and Bodywork Lab, is the opposite: about four times as many people come from Bing as from Google. People reach the two sites by completely different routes.
The devices differ as well. Two-thirds of the people who come to this site from Google are on smartphones. Almost all of the people who come to Mind and Bodywork Lab from Bing are on computers.
Third, generative AI systems read this site more than 27,000 times in three months. Yet only a handful of people came to the site after seeing an AI’s answer. AI reads a great deal, but that reading has not yet turned into visitors.
Leaning Toward Google, Writing for AI
This is the site that leads to Rolfing session bookings, and it is Google that brings people here. So I decided to lean this site toward Google Search.
At the same time, I put a lot of content into the articles, so that generative AI would cite them.
In fact, a fellow Rolfer told me that when they asked a generative AI about the ideas behind Rolfing, this site was cited in the answer.
What Changed
The Home Page
On the old home page, the top showed only photographs, with no text. Below them, the page introduced three features of Rolfing: it has an end (ten sessions), clients find a comfortable posture on their own, and it encourages psychological change through the body.
The new home page drops the three features. Instead, the Japanese home page now opens with the heading “Rolfing: ten-session bodywork that restores a body in harmony with gravity.” Below it come a short explanation of what Rolfing is and the location, length, and fee of the trial session.
The trial session takes place in Shibuya, Tokyo, and lasts 90 minutes: a 30-minute consultation and 60 minutes of bodywork. The fee is ¥10,000, or ¥15,000 for a session in English.
The reason for the change lies in the first finding. At 41st for “Rolfing,” Google sends almost no one. So the heading now includes the word Rolfing, and the explanation responds to the words people actually search with. What the three features used to convey is now carried by this explanation and by the Start Here page described below.
Putting the location, length, and fee at the very top comes from the second and third findings. Google is what brings the people who book, and generative AI reads this site often. What helps both people searching on Google and people asking generative AI is having the information needed to book gathered where it is easy to find. Whether people arrive from Google or learn about the site from an AI’s answer, they reach the same information. A short, self-contained explanation is also easy for generative AI to use directly in its answers.
The home page also brings together the idea of combining a PhD with bodily intelligence and the idea of the Recognition OS. The Recognition OS has four layers, Logic OS, Sensory OS, Dialogue OS, and Perspective OS, and Rolfing works with the Sensory OS. Readers can also start from three entry points: the structure of the body, the memory of the body, and the science of the body.
- Why Shoulder Pain and Low-Back Pain Are Not Healed by Massage? — The Science of Fascia, Posture, and Gravity
- Why Do Emotions Stay in the Body? — Understanding Rolfing Through the Body’s Memory — The Science of Fascia, the Autonomic Nervous System, and Trauma
- Is Rolfing Pseudoscience? — Understanding Rolfing Through the Science of the Body: The Latest Fascia Research and a PhD’s Perspective on the Evidence
Start Here
The old website had no way in for first-time visitors. A site with 540 articles is an accumulation of work for the person writing it, but for someone arriving for the first time, it is simply a place where it is unclear where to begin.
So I created a Start Here page at the far left of the menu. On the Japanese site, it offers four entry points based on interest and ten themes to read through.
Four entry points: Learning about Rolfing / The connection between body and emotion / Getting to know the practitioner / Trying a session
Ten themes:
- What is happening in the body: fascia / gravity / perception / energy work
- Where the practitioner is looking from: phenomenology
- How people changed after receiving sessions: yoga × Rolfing / client stories
- What science says: posture and movement science / somatic psychology / attachment theory
Of these, the pages on fascia, gravity, and perception, and the page on phenomenology under “Where the practitioner is looking from,” were written after I had commissioned the site. None of them was part of the original plan.
The Start Here page also gathers ten frequently asked questions: what Rolfing is, how it differs from massage and other bodywork, how many sessions are needed, whether it hurts, how to pay, sessions in English, and more. A question-and-answer format is easy for generative AI to use directly in its answers.
The page did not take this shape from the start. Partway through the build, it contained three separate ways of directing readers: four entry points by interest, a suggested reading order, and seven entry points by theme. They led to nearly the same places, divided up three different ways. The job of Start Here is to send each reader to the right page, and the more the ways of dividing overlap, the less well it does that job. So I narrowed them to two: the four entry points and the list of themes.
The list of themes also changed. It began as nine themes sorted along three axes: the person, experience, and science. I reorganized it into ten themes under four questions.
One reason was that the names of the axes reflected the writer’s way of sorting things. Labels such as “Getting to know Hidefumi Otsuka,” “Learning from those who have experienced and practiced it,” and “The science of the body, nervous system, and mind” did not let readers see themselves in them. So I turned them into questions, in the order readers tend to wonder: what is done to me, who does it, how will I change, and is there any evidence. Listing them as nouns would turn them back into category names, so they appear as questions.
The other reason was the number. I went from nine to ten to match the ten sessions of the Rolfing series. “Getting to know Hidefumi Otsuka” was removed and left to the “Getting to know the practitioner” entry point and the Profile menu. Fascia moved out of the science group to the top of “What is happening in the body,” and phenomenology became a group of its own.
The English site has its own Start Here page as well.
The Number of Articles
Before the launch, I reread fifteen years of articles one by one, sorting them into ones to keep, ones to let go, and ones to merge into a single piece. The Japanese articles went from about 540 before I began to 432. The English articles went from about 170 to 437 as I translated the Japanese ones. There are more in English because some articles exist only in English.
Two Sites, Two Roles
As the second finding showed, people reach this site through Google and Mind and Bodywork Lab through Bing. I divided their roles as well. This site covers the Sensory OS, which Rolfing works with, while Mind and Bodywork Lab explores the Logic, Dialogue, and Perspective OS. This site takes bookings, and Mind and Bodywork Lab is where I share ideas. Mind and Bodywork Lab’s strength in Bing also matters for being cited by ChatGPT and Copilot. From the home page and Start Here, readers can move on to Mind and Bodywork Lab’s latest articles and series.
The Profile and the English Site
The Profile menu traces how I went from a PhD to Certified Rolfer, along with the record of twelve years of training.
- From PhD (Medicine) to Certified Rolfer®
- The Path to Becoming a Rolfer — A 12-Year Training Record
- Why Are Advanced Rolfer Sessions “Profound”? — Three Differences, Ten Years After Rolfer Certification
Every page has a switch to English that leads to the corresponding English page.
Conclusion
In this post, I have written about what changed on the new website compared with the old one, why I changed it, and what evidence I based the changes on.
The site ranked 41st for “Rolfing.” People come here through Google. Generative AI reads the site often, but visitors are not yet following. From these three findings, I rebuilt the home page, created Start Here, and divided the roles of the two sites.
As before, I asked Tomonori Kirigishi, a digital life creator I have worked with for many years, to build the site. I am grateful to him.
First-time visitors can begin with Start Here.
This is a four-part series. The first three posts are here.