Six pieces. Written for this page, not for a client.
None of the work below was commissioned, delivered or anonymised. The businesses are invented and obviously so — a forecasting tool for corner shops, a linen brand, eleven flats in a Sofia building that does not exist.
Each piece says the same thing again where it sits, which is on purpose. On a page like this the label is the only thing standing between a sample and a claim, and one disclaimer at the top is a thing nobody reads twice. There are no numbers attached to any of it either: these were never shipped, so we have no idea what they would have done.
The part worth reading is the note underneath. The writing is the visible half; the decision behind it is what you would actually be buying.
Written to show method. Not delivered client work.
Positioning statement for an invented product: inventory forecasting for independent shops with one to five locations. Not homepage copy — the internal document the pages get written against.
for
Owner-operators of one to five shops who do the ordering themselves on a Sunday evening, with no buyer, no planner and nobody to ask.
against
Not the spreadsheet. The spreadsheet works, it is free, and the owner built it. Against the retail planning suites, which are priced and designed on the assumption that a head office exists.
claim
It orders the way somebody who has worked your shop floor for two years would order. It knows the Thursday before a long weekend is not a normal Thursday. It knows which lines you will forgive being out of stock, and which one you will not.
concession
It will not manage suppliers, it will not price anything, and for the first six weeks it will be worse at this than you are.
A positioning statement is mostly a set of refusals, and the useful one here is the last line. Naming a period where the product loses to the owner costs a claim and buys the only thing an owner-operator is deciding: whether the people who built it have ever stood behind a counter. The competitor choice is the other decision. Attacking the spreadsheet insults the buyer, because the buyer made the spreadsheet — so the statement concedes it and picks a fight with the enterprise suite instead, which the buyer already resents.
demonstrates brand positioning
Written to show method. Not delivered client work.
One pillar from a message hierarchy, for an invented B2B platform that schedules lorry arrivals at warehouse loading bays. The proof points sit under it. The line that got cut is shown too.
pillar
Drivers stop queueing in your yard.
proof
The slot gets booked by the haulier rather than assigned by your office, so the person who knows the route picks the time.
proof
A late arrival is re-slotted on its own, and the bay it lost goes to whoever is already parked outside.
proof
Your gatehouse sees the day as a list of names and times, on one screen, without logging into anything.
cut
Cut from this pillar: “full visibility across your entire supply chain.” True, useless, and the first line on three competitors’ homepages.
A pillar has to survive being repeated by a customer to a colleague, which disqualifies most of what gets written as one. “Drivers stop queueing in your yard” is short enough to be retold and specific enough to be wrong — a yard manager can walk outside and check. The proof points then descend by who is affected, from the haulier to the process to the man on the gate, so a reader who stops after the first one has still read the most persuasive. The cut line matters more than the three that stayed: keeping it would have lifted the pillar into category language, where it gets measured against sentences the buyer has already read this quarter.
demonstrates messaging architecture
Written to show method. Not delivered client work.
Winback email for an invented DTC brand selling linen bedding. Sent to someone who bought once, eighteen months ago, and has opened nothing since. This is the email where most brands send a discount code.
subject line
Is it holding up?
body
You bought a set of our sheets in March last year. Eighteen months is roughly the point where linen stops looking new and starts looking good, so this is a real question and not a way into an email: are they holding up?
If they are, you don’t need anything from us. Linen lasting is most of the point, and a brand selling you a second set this soon has either made something that wears out or talked you into something you didn’t need.
If they aren’t — thinning at the hem, a seam gone, patchy colour after washing — reply and say which. We’ll repair it, replace it, or tell you honestly what happened. Sometimes what happened is that we changed mills and the first run was worse.
Nothing to click, no code. We only want to know which of those two it is.
The decision is to write a winback that does not try to win the sale back. A product sold on durability cannot send a reorder prompt without arguing against itself, so the ask becomes information. Then the third paragraph admits a bad production run, which is the line the first two depend on — a brand that concedes one failure can be believed about the rest. A discount code would have been the easy version of this email and would teach the reader to wait for the next one.
demonstrates lifecycle & retention email
Written to show method. Not delivered client work.
The first screen of a sales page. Invented subject: a steel-framed bicycle built in Plovdiv and sold direct, at a price that needs a paragraph before anything else happens.
headline
The last bike you’ll buy for a decade, assuming you don’t leave it outside
opening
This bicycle costs about three times what a shop will sell you for the same commute, and for the first year you won’t be able to feel the difference.
The frame is steel. Steel is heavier than aluminium and it outlives it. Ours is brazed by two people in Plovdiv who were doing this before anybody called it heritage, and every joint can be repaired by a competent frame builder anywhere in Europe — which is not true of the carbon frame at the same price.
So this is not an argument about weight, and if weight is what you want, buy the carbon one. The argument is that in 2036 this will still be a bicycle. By then it will have needed two sets of wheels, some cables and one repaint.
A price like this has to meet the objection in the first sentence, not the twelfth. The opening is the reader’s own thought said out loud before they can think it, which is what earns the second paragraph. Two further decisions: the competitor named is carbon at the same money rather than the cheap commuter bike, because that is who the buyer is actually choosing between; and the claim is dated. “In 2036” does work that “built to last” cannot, because a year is checkable and slightly unsettling, and the maintenance list immediately after it stops the promise reading as a boast.
demonstrates sales page copy
Written to show method. Not delivered client work.
Opening of a release page for an invented development: eleven flats in a 1926 building in central Sofia, restored and sold rather than let. Written before a single floor plan gets shown.
opening
The building went up in 1926 as four large apartments, for one family and the people who worked for them. It then spent forty years divided into rooms with numbers on the doors. We have put it back to eleven.
Not four, because nobody lives that way now. Not the twenty-six the floor area would legally allow, because twenty-six is how you end up with a windowless corridor inside a building with three-metre ceilings.
Eleven, then. Every flat has at least one original window — the deep reveal, the brass catch that never quite closes in August. We kept those. We also put in wiring that won’t need doing again, a lift where the coal store was, and secondary glazing behind the original frames rather than instead of them.
Property copy defaults to the future: how you will live, what you will feel on the balcony. This one opens in the past and lets four numbers carry the argument — four apartments, forty years, eleven flats, and the twenty-six that were permitted and refused. A buyer at this level is trying to work out whether the developer had taste or only a spreadsheet, and the refused twenty-six answers that faster than any adjective available. The window catch that sticks in August is there for the same reason: a restoration that admits one flaw is easier to believe than one claiming none.
demonstrates launch copy
Written to show method. Not delivered client work.
Two pages from a voice guide, reduced to the part people use. Invented subject: three small mountain hotels under one owner, written for whoever posts, replies and answers the phone.
we say
It’s a two-hour drive and the last forty minutes are slow. Bring a coat, even in June.
we don’t say
Nestled in the heart of the mountains, our alpine sanctuary is the perfect escape for the discerning traveller.
why
Both sentences are trying to say remote. The first one is useful, and a guest who reads it and books anyway has chosen us knowing about the drive. The second is trying to be liked.
the rule underneath
When a guest asks something we would rather they hadn’t, answer it in the first sentence. The rooms have no air conditioning. The kitchen closes at nine. Children are welcome and there is nothing here for them.
never
Never “unfortunately”. Never “kindly note”. Never an exclamation mark in a reply to a complaint. Never “as per our policy”, which is a phrase that has already lost the guest by the third word.
Voice guides fail when they are lists of adjectives. Nobody can act on warm or authentic at eight in the morning with an awkward booking enquiry open, so every entry here is a pair — the sentence we would send, the sentence we wouldn’t, and the reason sitting under both. The rule about answering the unwelcome question first is the load-bearing entry, because it is the only part of the document that changes what a receptionist types today. That is the test for the whole guide: if it cannot be obeyed before coffee, it is decoration.
demonstrates brand voice guide
What a page of samples can’t tell you
Whether the argument is right. Every piece above was written with nobody in the room — no sales team to survive, no founder disagreeing about who the product is for, and invented research behind all of it. That is the easy version of this job.
The real one starts with interviews and a brief we argue with you about, and the sentences arrive last. Read these for the decisions. Read method for what actually happens, including what triggers hourly billing and what each of the 2 revision rounds is for.
Send a brief and we’ll argue with it
Tell us what you sell and which page you already suspect is the problem. You get a reply from a person within two business days — questions, and what we think the actual issue is. No charge for that part.