Rebuilding our own site meant facing the trap we sell partners out of.

The delay

Years afraid of our own advice

For years the old site said 'unleash your creativity' and meant nothing. We knew it was wrong. We did not fix it, because the honest way to fix it looked like the exact trap we sell partners out of. Hand it to developers. Lose the weeks to briefing and chasing. Watch the voice thin out in the handover. Become the person who has to approve everything, again. We had written that trap on the page we did not yet have. About the page itself, we were living in it.

We talked to several people who build sites. Each conversation ended in the same place: a management and delay rabbit hole we had neither the time nor the appetite for, and a quiet fear of getting lost in code until the voice was gone.

So we waited until two things were true at once. That the idea was too large to hand over without losing it. And that we could carry it ourselves without disappearing into work we do not do for a living. When both held, we built it.

On purpose

What we decided, and why

Compression, because the first version drowned

The first design held everything we wanted to say and was unusable, ours included. We needed to keep most of the information and lose some of the shape we were attached to. The need was practical before it was philosophical, and then it became the argument: attention is the scarce thing, so say the short version first and keep the long one underneath for whoever wants to check.

You keep the identity, we do the verbs

You own the voice, the standard, the head, the brand. We read, write, run, record, hold. We do not write 'our creative vision' over your work anywhere, because it is not ours. That was a rule from the first line.

Set the obvious reading, then replace it

Pages here tend to say: you thought it was this, it is actually that. At first we wrote that way without noticing. Then we saw it was doing the work, and made it a mechanism on purpose. It is how the pages think out loud instead of asserting.

The site performing its own service

This was the most deliberate decision of all. What we sell is taking a person's accumulated judgement, compressing it, keeping the character, and making it usable by others. So the site is that, done to ourselves: a compressed recommendation on top, the obsessive record underneath.

Arriving mid-process

The image on every main page is a timecode, something being worked on, not a finished object. We sell continuous production, not a shelf of endings, so the picture should say 'in progress'.

Keeping all of it

The casework carries every partner review, not a chosen six. A chosen six is a claim; the whole set is a record. The ones that are not five stars are still findable, on purpose.

Restraint in the machinery

No marketing tags. Analytics only after you choose, with accept and reject the same size. The back of the site behaves like the front. We do marketing; we would rather not behave like marketing people.

Bespoke effort only where judgement lives

Webflow for the boring, reliable parts. Custom work saved for the few places it changes what you understand, a demonstration you can feel rather than a decoration that announces itself. It is the same rule we use at the bench: do not rebuild the boring infrastructure, spend the attention where taste actually decides something.

Not on purpose

What emerged, and happened to fit

Not all of it was planned. The clearest example was a conclusion, not a design. You start reading, it holds you, then you remember you have work to do and you leave, and the page has just demonstrated its own first claim back to you: your attention is the scarce thing, not your hours. We did not author that. We noticed it afterwards and left it alone, because it was true. We are telling you it was not authored so you can trust the parts that were.

Some readings of the site we find plausible but will not claim as intent. The old-world imagery, the chess, the meanings people read into them. Some of it is deliberate. Some of it is just margins. We would rather say which is which than let every twelve-pixel gap sound like philosophy.

The honest limit

What we cut, rushed, and will not claim

The first design cost us pages we liked. Compression was the right trade and it still stings a little.

The partner area is not here: one screen where everything we keep about a partner, the reports, the deliverables, the status, sits in one place. It is the ambition we want most and have not built, because it is roughly the whole site again. Postponed, not abandoned.

Insights became guides. There were too many, and one day I could not find a thing I was sure I had written, or half-written with someone. If the author cannot navigate it, the reader has no chance. Guides are the correction, and by my own eye they are not finished.

Some of this shipped rushed, because we set a launch date and we keep launch dates, including the ones we give ourselves. A few ideas were judged impractical and dropped. A few will arrive after launch. A few turned into other things on the way.

And the honest limit. This does not prove we are good at the work. It cannot. It is a demonstration, not a sample: it shows we can build what we describe, not that the work itself is good, which is shown by action, in a partnership. What this shows is narrower and, to us, worth more than a reel: with nobody else involved, and tools that only make it easier rather than easy, we took what we think and made it stand up. Whether we then make your channel work is a thing you test, not a thing you read.

Source notes

The shoot, and what a stranger can name

The strategy image is a chess game, and not an ordinary one: a single frozen position, Fischer against Tal, the twenty-second move of the twenty-seventh round. A line long attributed to Tal names the place he took opponents: a deep dark forest, where two and two make five and the way out is wide enough for one. The site is meant to be that, should you accept the challenge: go in and it complicates, pleasantly, until you find your own way out. We take no side. A game holds both the player who set out to be champion and the player nobody remembers by name, only by the move; both are in a person, and in the two of us. The idea came from agadmator's chess channel, which we like and have no relation to.

The move itself is a small story about nerve, left for whoever digs: Fischer wrote down the strongest move, then unwrote it because Tal grinned across the table, and played a weaker one. Read it however you like.

The rest is inspectable from the outside. It is a Webflow site, served from Webflow's own files. The heroes are one repeating component, a timecode, on purpose. It remembers which guides you have read, once you allow it. A few case studies carry small interactive pieces built to make an analytical point perceptible, a microphone out of phase, a timeline, a chord, rather than to show that someone owns an animation library. Services, insights and casework each run through one shape, one set of fields, many pages: define the standard once, feed different material through it. The same doctrine as the bench.

There is little here to take on faith, because almost nothing is hidden. This is our own record of the build, with no partner's data in it, so nothing is withheld. The few facts it rests on, the reviews we kept, the tools we used, the day it went live, sit elsewhere on the site with their dates. To check any of it, open the page it describes and look.

What you would get

Test us, and we will test you

Read the site as a stranger would, then send us something real. The first conversation is free, and if we are the wrong fit we will say so before anyone talks about money. Request yours.