01 Problem
The problem
The studio's first site existed to send visitors to a packaged product with one fixed price, and the feedback on it was that it looked like the AI-generated landing page.
When it moved into the product's own codebase the problems multiplied: a navigation that asked strangers to pick one of three worlds, two of which had nothing to sell; two heroes a screen apart; ten sections restating the journey; an ending that restated the positioning; and four calls to action pointing at a product I had retired.
02 Evidence
How I knew it was real
- Of the three worlds in the navigation, two had nothing to sell.
- Measured on a 390 by 844 phone, the only button at the end of the page landed behind the cookie bar. A first-time visitor on a phone reached the end of the journey and was offered nothing.
- After I retired the product the site sold, four of its buttons kept pointing at it for fifteen days.
03 Decision
What I set out to change
One company, one argument, one way out: show a business running itself, name the work it takes off the owner, and end on a single button.
What I deliberately left alone
- Search reach from the homepage. The page is one journey now, and its own keyword surface went with the sections - a cost I wrote down rather than hid.
- Prices. There is no packaged offer to buy; every call to action leads to one enquiry form.
The choices worth arguing with
Chose One company, one journey, nothing under it.
Over Three worlds - Business, Body and Mind - to choose between in the navigation.
Because Asking a visitor to pick a world before they know what the company is, is homework.
Chose An ending that names the work it takes off the owner: three problems and one button.
Over An ending that restated the positioning - a living operating system, three pillars, a loop.
Because Nobody buys what a thing is; they buy the Tuesday it gives back.
Chose Every call to action leads to one form.
Over Buttons into a packaged product with a fixed price.
Because The work is bespoke now, and a button that lands a warm visitor on a retired page is worse than no button at all.
Chose The phone layout is trimmed against a measurement, not a guess.
Over Shrinking everything until the ending fits.
Because The cookie bar is pinned to the bottom, so the fix had to be a shift, not a diet: on short phones the whole ending moves up 2.75rem.
04 Build
What I built
The homepage is one fourteen-viewport journey drawn in WebGL: a small company running itself, the traces of its processes, the moment one decision needs its owner, and then daylight - the three problems an owner recognises fastest, said plainly, and one filled button to the systems page, a page on this site.
There is nothing under the journey. The retired worlds redirect to the homepage and left the sitemap at the same time, their copy stays in the history in case it is ever wanted, and every link to the retired product was swept.
The phone is designed against measurements. On short screens the three problems keep their names and lose their explanations - the names are the argument - and the ending moves up so the button clears the cookie bar.
- SourceA stranger arrivesFrom a search, a pitch or a link.
- StepA company runs itselfFourteen viewports: its processes, its log, its clock.
- A personOne decision needs its ownerThe moment the story turns on.
- StepWhat it takes off youThree problems, by name.
- OutOne buttonTo the systems page, on this site.
- OutOne formWhere every call to action ends.
05 Validation
How I tried to break it
The only button at the end, on a 390 by 844 phone with the cookie bar up
Clears the bar. The problem explanations come off on phones, and the ending shifts up on short screens.
Broke, then fixed · 30 Aug 2026
The same ending at 1440, 768, 390 and 360 wide
Measured at each; the button clears the cookie bar at all four.
Held
Links to the retired product
91 links swept, and none reach the retired host.
Broke, then fixed · 30 Aug 2026
Old links to the three worlds
Redirect home, and left the sitemap - a sitemap full of redirects is one crawlers learn to distrust.
Broke, then fixed · 15 Aug 2026
The wordmark on the dark navigation bar
Was invisible, and is not any more.
Broke, then fixed · 30 Aug 2026
What did not work
- The first rebuild ported the look before the argument: three worlds, two heroes, ten sections. It took several rounds of cutting to reach one journey.
- I retired the product the site sold, and four buttons kept pointing at it for fifteen days.
- The homepage's search description kept promising a fixed price after the rest of the site had stopped. I found it while writing this case, and it is fixed in the same change.
06 Outcome
What changed
- 3 to 1 worlds in the navigation
- Business, Body and Mind retired on 2026-08-15; their pages redirect home.
- 10 homepage sections removed
- Cut on 2026-08-15, leaving the journey alone on the page.
- 91 links swept to the retired product
- Swept on 2026-08-30, when the last buttons to it were removed. None remain.
- 4 screen widths the ending is checked at
- 1440, 768, 390 and 360 pixels, against the cookie bar, 2026-08-30.
The site says one thing, in one scroll, and ends on one button - to a page on this site, not to a product that no longer exists. Whatever a visitor wants, they arrive at the same form.
What I cannot tell you yet is whether it converts better. Analytics are wired and wait for consent, and no before-and-after numbers exist in anything I can check.
What I measure next
- Where visitors stop in the journeyScroll depth per viewport - not recorded today.
- Enquiries per visit, before and afterThe enquiry endpoint and consented analytics, compared across the change.
- Search visibility the sections took with themSearch Console queries for the homepage, against the keyword surface that was removed.
07 Learning
What I learned
Nobody buys what a thing is; they buy the Tuesday it gives back. So the ending names three problems instead of describing an architecture.
Serving an old link is not the same as advertising one. Links already in people's inboxes still work; nothing on the site points at them.
A phone layout is a measurement problem. The cookie bar was the one element on the page nobody had designed, and it decided whether the ending worked.
What I would change next
- Record scroll depth, so the journey's drop-off becomes a number.
- Give the homepage back some search surface without putting a section stack under the journey.