One Brand three readers A single source of truth
We make brands readable to the person who skims, the model that retells, and the agent that acts. We start with a number.
Why this matters to your sales
Your buyers already ask machines.
Three readers now decide every purchase: the person who skims, the model that retells you to buyers who never visit, and the agent that acts only on what it can verify.
Conversion for the top fifth of Shopify stores, against a 1.4 to 1.8% average. That gap is usually fixable words.
Shopify benchmark, 2026
Year-on-year growth in AI-referred sessions to Shopify storefronts, Q1 2026.
Blackbelt Commerce, 2026
Of B2B buyers now use generative AI during the purchase cycle.
6Sense, 2025
Projected United States agentic-commerce sales by 2030: agents that act on your words, or on a rival’s.
Bain, 2026 ($300 to 500B)
the reading
One estate, three contracts
Every page is read three ways
Each sentence pays for the next, or the reading ends.
She owes you nothing at 23:40, eleven tabs open, already leaving. The first sentence exists to earn the second.
The first line answers the question she walked in with, in its first breath
Claims walk in with their receipts, or wait outside
The larger the number, the quieter the sentence that carries it
Any flourish that can be cut without consequence is cut
Every receipt is a passport. Everything else is dead weight at the border.
It reads like a clerk with no patience for charm, in order to retell you to a buyer you will never meet.
The answer sits whole, in the first breath of the page that owns it
Who you are is told once, canonically; every page defers to that telling
Facts agree with each other everywhere, because the clerk cross-checks
Dates swept like a shopfront: freshness is the cheapest proof of life
The brand that shortens the worker’s task gets the worker’s recommendation, at four in the morning.
A worker with a task and a checklist, built by engineers who trust nothing a website says about itself.
The five documents that matter, handed over inside thirty seconds
The code beneath the page says exactly what the English says
Score, method and price answered from the published record
Every claim written so a checking system can fail it
Services & pricing
The work sets the price.
Get your number
Free check. Three readers score your site out of 100.
Read the why
The Diagnostic goes deeper: eight areas, proof for every point, editor-signed.
Make the fix
Use the Toolkit yourself, or let us build your rules with Studio Start. About a week.
↻ touch each card to turn around
The Free Score
$0
Where you stand today, by all three readers.
The door · free
The Free Score
Your score out of 100
One verdict per reader: person, model, agent
The #1 fix that moves your score most
The Diagnostic
$1,000
Not where each reader stands — why. Which pages lose the sale.
Direct entry · no prerequisites
The Diagnostic
All 8 surfaces audited, with evidence
Page-by-page fix list, sized as tasks
Your key message shown 3 ways (person/model/agent)
Editor’s signature + 1 clarification round
The Readability Toolkit
$2,500
$1,500 after a Diagnostic
Your team does the fixing, with the house’s own instruments.Buy alone, or after the Diagnostic
The Readability Toolkit
9 house instruments + worked example
Live session building one offer with an editor
Every file, yours to reuse
Studio Start
$5,000
$4,000 after a Diagnostic
One named editor rebuilds your voice from your own documents.Requires the Diagnostic to apply
Studio Start
Voice rules, drawn from your own lines
Usable by people and by tools (stays on-brand)
Handover walkthrough + 30-day review
↻ touch each card to turn around
The shift
Rented vs owned
SEO was
Being found
Tuned to one reader’s passing quirks. Decays on the model-update schedule. The fee is certain, the decay is scheduled, and the schedule belongs to someone else
Readability is
Being carried
Writing and evidence good enough that a person acts on it, a model retells it faithfully, and an agent completes a task on it. Substance is tuned to all readers, including the ones that do not exist yet.
1/3
What we hold
Principle
How it’s kept
Truth before reach
If a claim lacks evidence, we reduce or remove
the claim. We would rather say less and prove it.
Human first
If a machine-facing change makes a page worse
for people, we reject it.
Receipts over adjectives
Proof, dates, examples and mechanisms over superlatives, everywhere.
One canonical story
A price that argues with itself is a brand that does.
We resolve contradictions first.
Transparency,
no black box
The score is published and versioned; you can see why a point was won or lost.
Honesty about limits
We promise the work, not the downstream outcome. No guaranteed rankings, citations, traffic or leads.
2/3
First light
The house, on itself
A brand can rent noise. It cannot rent the ability to be carried.
The estate is the asset. The engine is a replaceable reader.
They measure. We write. Measurement is the door;
the work is the house.
The system’s first client is its author. Our estate runs the full system on itself, and one line on this page is deliberately blank: the date of our own first light, the first verified, faithful answer carrying this house. The mechanism is built and the date is not ours to type. It stays blank until the readers fill it.
3/3
The Readability Manifesto
Being found was the old craft.
Being carried is the new one.
The basis of everything this house does.
How far does your voice carry?
Get your free scoreOne number, three verdicts, and the single repair that would move your score most.
Read by instruments, banded honestly, delivered to your inbox.
What we will not sell you
We do not sell rankings or mentions, and we make no promise that any engine will cite you. We measure and improve whether your brand can be read, retold and acted on, with evidence, and we put a published, versioned instrument behind every number.
The promise is the work, honestly scored; the outcome belongs to the readers.
THE DOOR
Who’s reading your website?
Our instruments read your pages against the eight published checks.
Your score arrives by email: one verdict per reader, and the single repair that would move it most. Two fields, about a minute.
The honesty block
The Brand Readability Score is free, complete on its own terms, and enough for most companies to act on. The Diagnostic exists for companies that need the full evidenced reading; it is not a required next step. A company that acts on the free score alone has used it correctly.
The score has its own page,
with the Door, the method in brief,
and the limits printed plainly.
Free. No account, no card. One reading per website per quarter.
The instrument
Eight surfaces. Three readers. One number you can check, dispute, and improve.
The Person
reads you
The Model
retells you
The Agent
acts on you
One estate, three contracts
Every page is read three ways
Each sentence pays for the next, or the reading ends.
She owes you nothing at 23:40, eleven tabs open, already leaving. The first sentence exists to earn the second.
The first line answers the question she walked in with, in its first breath
Claims walk in with their receipts, or wait outside
The larger the number, the quieter the sentence that carries it
Any flourish that can be cut without consequence is cut
Every receipt is a passport. Everything else is dead weight at the border.
It reads like a clerk with no patience for charm, in order to retell you to a buyer you will never meet.
The answer sits whole, in the first breath of the page that owns it
Who you are is told once, canonically; every page defers to that telling
Facts agree with each other everywhere, because the clerk cross-checks
Dates swept like a shopfront: freshness is the cheapest proof of life
The brand that shortens the worker’s task gets the worker’s recommendation, at four in the morning.
A worker with a task and a checklist, built by engineers who trust nothing a website says about itself.
The five documents that matter, handed over inside thirty seconds
The code beneath the page says exactly what the English says
Score, method and price answered from the published record
Every claim written so a checking system can fail it
What we will not sell you
We do not sell rankings or mentions, and we make no promise that any engine will cite you. We measure and improve whether your brand can be read, retold and acted on, with evidence, and we put a published, versioned instrument behind every number.
The promise is the work, honestly scored; the outcome belongs to the readers.
Services & pricing
The work sets the price.
Get your number
Free check. Three readers score your site out of 100.
Read the why
The Diagnostic goes deeper: eight areas, proof for every point, editor-signed.
Make the fix
Use the Toolkit yourself, or let us build your rules with Studio Start. About a week.
↻ touch each card to turn around
The Free Score
$0
Where you stand today, by all three readers.
The door · free
The Free Score
Your score out of 100
One verdict per reader: person, model, agent
The #1 fix that moves your score most
The Diagnostic
$1,000
Not where each reader stands — why. Which pages lose the sale.
Direct entry · no prerequisites
The Diagnostic
All 8 surfaces audited, with evidence
Page-by-page fix list, sized as tasks
Your key message shown 3 ways (person/model/agent)
Editor’s signature + 1 clarification round
The Readability Toolkit
$2,500
$1,500 after a Diagnostic
Your team does the fixing, with the house’s own instruments.Buy alone, or after the Diagnostic
The Readability Toolkit
9 house instruments + worked example
Live session building one offer with an editor
Every file, yours to reuse
Studio Start
$5,000
$4,000 after a Diagnostic
One named editor rebuilds your voice from your own documents.Requires the Diagnostic to apply
Studio Start
Voice rules, drawn from your own lines
Usable by people and by tools (stays on-brand)
Handover walkthrough + 30-day review
↻ touch each card to turn around
The visibility dashboards
Track how often AI mentions you
Benchmark citations and sentiment against rivals
Show you the gap, then stop
Leave the writing, the part that changes the answer, to you
The Writing House
Reads your pages the way a person, a model and an agent each read them
Names where you are misread, with the evidence beside every finding
Does the writing that changes what the model and the agent read
Starts with a free score that is complete on its own terms
We do not penalise a buyer for starting where they should
Run the Diagnostic once and its $1,000 carries into the single next product you choose, held for ninety days. The Diagnostic is step one of the larger work: its exact number and fix list are where the Toolkit and Studio Start begin, so its fee carries with you.
The payoff, measured
What readable brands earn
By the measurement industry’s own 2026 counts, the brands the answers carry are paid in clicks, conversion and recognition.
Organic clicks
Brands cited in AI answers earn a third more organic clicks than uncited rivals on identical queries.
For your revenue
More qualified visits from the same demand, without more ad spend.
Paid clicks
The same cited brands earn nearly double the paid clicks, because the answer has already made the introduction.
For your brand awareness
The machine retells you by name, in rooms you never enter.
Conversion rate
Buyers who arrive from an AI answer convert at a multiple of the old rate: the answer pre-qualified them before your door opened.
For your conversion
Fewer, warmer visitors who already believe the claim your page must keep.
How far does your voice carry?
Get your free scoreOne number, three verdicts, and the single repair that would move your score most.
Read by instruments, banded honestly, delivered to your inbox.
The shift
Rented vs owned
SEO was
Being found
Tuned to one reader’s passing quirks. Decays on the model-update schedule. The fee is certain, the decay is scheduled, and the schedule belongs to someone else
Readability is
Being carried
Writing and evidence good enough that a person acts on it, a model retells it faithfully, and an agent completes a task on it. Substance is tuned to all readers, including the ones that do not exist yet.
1/3
What we hold
Principle
How it’s kept
Truth before reach
If a claim lacks evidence, we reduce or remove
the claim. We would rather say less and prove it.
Human first
If a machine-facing change makes a page worse
for people, we reject it.
Receipts over adjectives
Proof, dates, examples and mechanisms over superlatives, everywhere.
One canonical story
A price that argues with itself is a brand that does.
We resolve contradictions first.
Transparency,
no black box
The score is published and versioned; you can see why a point was won or lost.
Honesty about limits
We promise the work, not the downstream outcome. No guaranteed rankings, citations, traffic or leads.
2/3
First light
The house, on itself
A brand can rent noise. It cannot rent the ability to be carried.
The estate is the asset. The engine is a replaceable reader.
They measure. We write. Measurement is the door;
the work is the house.
The system’s first client is its author. Our estate runs the full system on itself, and one line on this page is deliberately blank: the date of our own first light, the first verified, faithful answer carrying this house. The mechanism is built and the date is not ours to type. It stays blank until the readers fill it.
3/3
The Readability Manifesto
Being found was the old craft.
Being carried is the new one.
The basis of everything this house does.
One estate, three contracts
Every page is read three ways
Each sentence pays for the next, or the reading ends.
She owes you nothing at 23:40, eleven tabs open, already leaving. The first sentence exists to earn the second.
The first line answers the question she walked in with, in its first breath
Claims walk in with their receipts, or wait outside
The larger the number, the quieter the sentence that carries it
Any flourish that can be cut without consequence is cut
Every receipt is a passport. Everything else is dead weight at the border.
It reads like a clerk with no patience for charm, in order to retell you to a buyer you will never meet.
The answer sits whole, in the first breath of the page that owns it
Who you are is told once, canonically; every page defers to that telling
Facts agree with each other everywhere, because the clerk cross-checks
Dates swept like a shopfront: freshness is the cheapest proof of life
The brand that shortens the worker’s task gets the worker’s recommendation, at four in the morning.
A worker with a task and a checklist, built by engineers who trust nothing a website says about itself.
The five documents that matter, handed over inside thirty seconds
The code beneath the page says exactly what the English says
Score, method and price answered from the published record
Every claim written so a checking system can fail it
the reading
How far does your voice carry?
Get your free scoreOne number, three verdicts, and the single repair that would move your score most.
Read by instruments, banded honestly, delivered to your inbox.
Who it serves
Every path here leads to the same word:
misread
Wherever you sit, the cost is the same, and the fix waits right below
Touch to open
How far does your voice carry?
Get your free scoreOne number, three verdicts, and the single repair that would move your score most.
Read by instruments, banded honestly, delivered to your inbox.
Notes on being read.
What we learn making brands legible to the three readers — the person who skims, the model that retells, the agent that acts.
Three readers, one source of truth
The Writing House · method
Essay
Three readers, one source of truth
Why one page has to satisfy a skimmer, a model and a machine at once.
FIELD NOTE · THE SCORE
homepages passed the model read
Field Note
We scored 40 homepages. Only six passed
Where storefronts break for a reader that reads before it buys.
“The agent never sees your hero image.”
CONVERSATION · Nº 03
Conversation
“The agent never sees your hero image”
A working session on what actually reaches an automated buyer.
FIELD NOTE · CONVERSION
top-fifth stores convert here
Field Note
The gap is usually fixable words
The top fifth convert at 3.1%. Most of the time the words are the difference.
What a machine reads first
The Writing House · craft
Essay
What a machine reads first
The order of the page is the order the model trusts you in.
DISPATCH · AI TRAFFIC
growth in agent-referred sessions
Dispatch
Your buyers already ask machines
Agent-referred sessions grew 8× in a year. Here is who they are.
“Read it back the way a machine would.”
CONVERSATION · Nº 04
Conversation
Reading a homepage out loud
What breaks when a model narrates your storefront back to you.
The skimmer and the agent
The Writing House · readers
Essay
The skimmer and the agent, side by side
Two readers who never scrolled, and what each one took from line one.
DISPATCH · SIGNAL
is all a model spends on you
Dispatch
A model spends 1.2 seconds on you
What a machine decides in the time a person clears their throat.
No notes match that yet.
Field Note
The gap is usually fixable words
FIELD NOTE · THE SCORE
homepages passed the model read
The top fifth of Shopify stores convert at about 3.1%. The average sits between 1.4 and 1.8. That gap is rarely a design problem or a traffic problem — most of the time it is the words, and the words are fixable in an afternoon.
We read forty storefronts this month the way the three readers do. The person who skims. The model that retells you to a buyer who never visits. The agent that acts only on what it can verify. The same three sentences broke for all three, on almost every page.
6 of 40
homepages passed the model read, unedited
What the passing pages had in common
They named the thing plainly before they sold it. They put the one number that matters where a machine could find it. And they wrote the promise once, in one place, so all three readers left with the same sentence.
Being found was the old craft. Being carried is the new one.
None of this needs a redesign. It needs an editor — someone who reads the page as each of the three readers will, and cuts until the three of them agree.
Where The Writing House fits
We start with a number: the free score, which reads your page the way a model does. Then we do the editing — the afternoon's work that moves the sentences into the order all three readers trust. You can start with your score, or read how the method works.
Sources
Shopify commerce benchmark, 2026 · Blackbelt Commerce, AI-referred sessions Q1 2026 · The Writing House homepage review (n=40), Aug 2026 · Internal scoring rubric v3.
Continue reading.
The House · 3 notesThree readers, one source of truth
The Writing House · method
Essay
Three readers, one source of truth
Why one page has to satisfy a skimmer, a model and a machine.
“The agent never sees your hero image.”
CONV · Nº 03
Conversation
“The agent never sees your hero image”
A working session on what reaches an automated buyer.
DISPATCH · AI TRAFFIC
growth in agent-referred sessions
Dispatch
Your buyers already ask machines
Agent-referred sessions grew 8× in a year. Here is who they are.