Now

I have eleven retainer clients right now, against a ceiling of thirty. That ceiling is the whole point, and I'm not in a hurry to fill it. Past thirty the service degrades, the weekly rhythm slips, and the thing I'm selling stops being true. So eleven isn't a number I'm anxious about. It's eleven businesses whose search visibility I'm responsible for, every week, no exceptions.

The retainer is a real cadence, not a status email. Every client gets work every week, and once a month a report that stops and reads the dials — where the rankings actually moved, and what the next month is for. That's the consultancy, and it runs from We Build Stores.

StoreBuilder is the same craft sold a different way. The retainer is hand-built and capped; StoreBuilder is the pay-monthly version — £49 a month, or £995 to own it outright — for the businesses that want a proper site before they want a weekly relationship.

The launch product is a catalogue of twenty-two trade templates, each an Astro site with its own checkout. Sixteen of them ship with a keyword-matched .co.uk domain already attached; for the other six you bring your own. It's the same build and the same stack as the hand-made work, set down in a form a small business can pick up on its own.

I operate two stores. Indoor Luxury is live and running — somewhere north of five hundred products, pruned and curated rather than piled up. It's the one that proves the model: high-value goods that ship without an installer, sold to people who can afford them, run by one operator and the system instead of a warehouse and a team.

Garden Luxury is the one in build. No catalogue yet, and that's deliberate — the thesis there is to earn the search position first and let the supply arrangements follow the rank, rather than the other way round. There are a handful of other luxury domains I hold with checkout wired but no store behind them. They're held, not running, and I won't pretend otherwise. A registered domain is not a shop.

The operating system underneath all of this is where a lot of the month went, and three threads are worth telling you about.

The first is the system auditing itself. It reads its own codebase, finds dead code and stale documentation, and cuts it. The headline this month was removing an archived graveyard of a hundred and fifty-five dead command files — roughly forty-four per cent of the command code — with everything still passing afterwards. Along the way it corrected a health signal that had been quietly lying about how much of itself was actually wired. A system that prunes itself is worth more to me than one that only ever grows.

The second is syndication. When a piece of writing ships, the system now drafts the social posts off it, in voice, with a check that screens out slop — and then it stops and waits for me to approve. Nothing posts on its own. There's a hard rule that a post drafted for one identity never crosses onto another, because the practitioner voice and the commercial voice are not interchangeable, and I won't let them blur.

The third is housekeeping with teeth: a written house style guide for the sites I own, tooling that enforces it across them, and a separate weekly discipline that keeps the books reconciled rather than letting them drift into a monthly panic. None of it is glamorous. All of it compounds.

There are thirty-six published pieces in the collection now, and the cadence is two a week, Tuesday and Friday, committed a long way out. They're practitioner pieces only. I write from inside the work, not above it.

The argument has moved. It started with the methodology — the plain mechanics of how I actually work — then opened out into the philosophy of working with the system rather than instructing it, and now it's sitting in business-architecture and position-piece territory. The most recent one, The Nest, Not the Magpie, makes the case that what you take from reading other people's code is a function of the substrate you've already built, not the brilliance of the source. The few before it were about why operators burn their context in three turns, why context discipline is the entire skill, and the structural gap where governments collect the data but never build the front end. The pieces are a running record of the position as it develops, not a content calendar.

The thing genuinely unresolved this month is recognition across the answer engines. The old way to check whether a business shows up when an answer engine is asked is to ask one of them, once — and that's a lie dressed as a measurement. So I've been building an instrument that asks cold, asks several engines, and declares its own coverage honestly.

The part I keep turning over is that the measurement is the easy half. The real question is how a business earns recognition across these engines in the first place, not how cleverly you count it. I decided against paying for an automated path through an extra engine and kept a leaner, straighter setup instead. That's where my thinking is right now, and it isn't finished. It's the kind of question I'd rather sit with than answer too quickly.