Grow My Therapy
Grow My TherapyGrow My Therapy — done-for-you blog writing
We write the articles that bring people to your practice — the ones Google ranks, and the ones ChatGPT quotes when someone asks it who to see. Every post goes out sounding like you, because it was built from your words. And every post is held to the same rules your board holds you to, before you ever read it.
Most people who call me don't use the word anxiety. They say they haven't slept properly since March, or they snapped at their kid again over nothing, or they've read the same paragraph four times and still couldn't tell you what it said. That's usually where we start — not with a label, with the Tuesday that went sideways.
Voice pass applied · matched against your own writing sample · opens on something concrete · no diagnosis of the reader · no outcome promised · every fact, link and citation carried through unchanged
Why this matters now
For fifteen years there was one way people found a therapist online. That quietly became two, and the second one is growing fast. A practice that only shows up in the first room is already losing referrals it will never hear about.
Someone types that, or does EMDR actually work, and reads for twenty minutes before they call anyone. Google doesn’t rank practices — it ranks answers. A site with one page about your services competes for nothing. A site with two dozen articles that genuinely answer what people are asking becomes the site Google sends them to, and the person arrives already half-decided.
It’s also how you get found for the work you actually want. Broad terms belong to the directories. The specific ones — the modality, the population, the exact worry — are winnable, and they bring better-fit clients.
Most trauma-focused work begins with stabilisation rather than the memory itself 1. Sessions typically start with grounding skills you can use between appointments, and the pace is set by you 2 — there is no requirement to tell the whole story in the first hour 3.
They open ChatGPT and type exactly that, and get one answer built from a handful of sources. There is no page two. Google now does the same thing at the top of its own results. That answer either draws on your writing and names your practice, or it draws on somebody else’s.
Getting quoted in that answer is a different skill from ranking, and almost nobody is doing it yet for therapy practices. It has a name — AI visibility, or AEO/GEO — and it’s built into every post we write for you. How that works ↓
The honest objections
Nobody’s first concern is traffic. It’s that something will go out under their name that isn’t theirs. These four come up in nearly every call, and they’re all reasonable. Each one is answered by a specific part of how we work — not by a promise.
It’s going to sound like AI.
It would, if the draft were the deliverable. It isn’t. Every article goes through a separate voice pass built from a sample of your own writing — your sentence lengths, your openings, your habits, the words you use and the ones you’d never say. Your sample outranks every generic rule about “good writing.” The pass may change how it reads; it may not change a single fact.
→ Step 05, the voice pass
It’ll say something I don’t believe.
Before anything is written, we build a profile of your actual positions — which modalities you stand behind, how strongly you’ll state what the evidence shows, what you refuse to claim, which topics are off-limits for your practice entirely. Statistics are traced back to their source and confirmed. If a claim can’t be supported, it doesn’t appear.
→ Steps 01 and 04
It’ll say something my board would care about.
No diagnosing the reader. No promising or implying outcomes. No testimonials, no “best” or “#1,” no medication guidance, no real client stories. Crisis topics get visible resources and safe-messaging language. These aren’t guidelines we hope the writing follows — they’re checks that run on the finished article, every time.
→ The 19 checks
It’ll be generic. Anyone could have written it.
The brief for every post requires one thing a reader could not have guessed from the headline — a real distinction, a mechanism, something that could only come from someone who does this work. Openings must start somewhere concrete: a moment, a question you actually get asked. Reviewing that is one of nine scored dimensions, and generic writing fails it.
→ Dimension 7, clinical read
How a post gets made
This is the whole process, in order. It runs the same way for every article. The parts that need you take about an hour, once — at the start, and then only when you have a note.
A 96-question intake across 15 sections: who you treat and who you don’t, your modalities, your scope and licensure, your terminology, the claims you will and won’t make, the topics that are green, amber, and forbidden for your practice.
You don’t start from a blank form. We fill it in from your website first and mark every answer that was our guess, so you’re correcting a draft rather than writing an essay between sessions.
Produces your practice profile · every later stage reads from it
Separately from what you believe, we capture how you sound. You send a sample of your own writing — a newsletter, an old post, a page you wrote yourself. From it and your answers we build a voice profile: register, reading level, sentence rhythm, contraction habits, signature phrases, words to love, words to ban, whether you say client or patient, person-first or identity-first, and how your credentials may appear.
Your writing sample outranks every generic style rule in the system
The draft is written by the most capable model on the market, running a full editorial pipeline rather than a single prompt: it picks a template fit to the topic, builds a keyword-informed outline against what’s already ranking, dispatches a dedicated research agent to gather sourced statistics, writes answer-first so the point of each section lands in the first sentence, generates original charts and images, and places internal links.
Links are drawn only from a live map of your real site pages. The system cannot invent a URL, and anchor text is checked for length, description and repetition. The draft doesn’t advance until it scores 90 or above on the pipeline’s own 100-point rubric, through five blocking gates.
5-gate delivery contract · 90/100 minimum · 6–8 sourced statistics · hero and inline images with alt text
Fabricated statistics are the single most dangerous thing about AI writing in a clinical field, so we don’t trust the draft’s own citations. Each load-bearing claim is extracted, its cited URL is fetched, and the page is checked for whether the claimed figure is actually there. Matches are scored — exact, paraphrased, or not found. Anything unsupported is cut or replaced. Uncited claims are flagged rather than assumed.
Sources held to tier 1–3 only: peer-reviewed, government, established institution
A second, independent pass rewrites the prose to sound like you. It reads only your writing sample and your voice profile — deliberately not the research or the outline — so its whole attention is on cadence and word choice. It removes the tells: the hollow opener, the throat-clearing phrase, the rhythm where every sentence is the same length.
What it may not touch is fixed: facts, statistics, names, dates, citations, links, the disclaimer, the crisis box, and any labeled composite example all pass through untouched. It changes the writing, never the substance.
Nothing may be invented · structure and evidence preserved exactly
Then the article is measured — not reviewed, measured. Nineteen checks run against the finished file and produce a pass or fail with the evidence attached. They’re listed in full below. They cost nothing to run, so they run on every single post, and they catch the boring, expensive mistakes before a person ever spends attention on them.
Recorded per article, so a claim about quality is always evidenced
The mechanical checks answer “did the pipeline do what it was told?” They can’t answer the question you’d actually ask: would I put my license behind this? That’s scored separately across nine dimensions, against your profile rather than a generic idea of good therapy writing. Any finding must quote the article verbatim — an unquoted objection is an opinion, not a finding.
Nine dimensions · hard gate on safety and scope, regardless of total
Posts land in your site’s own dashboard as drafts, with images, formatting, structured data and meta tags already in place. You read it, you publish it. Practices that want us to publish directly can turn that on, and it’s recorded for that one practice only — there is no global switch that could take everyone live at once.
WordPress-native · draft by default, permanently, until you change it
This is the part that compounds, and it’s the reason the fourth month reads better than the first. Every published post arrives with a feedback link — one tap on your phone, no password. Write what’s wrong in whatever words you’d use on a call.
That note doesn’t get pasted into the next brief. It’s turned into a rule and written into your practice profile, above everything else in it, where it governs every future article. A phrase you strike is matched literally against every draft from then on, and a check fails loudly if it ever comes back. You will never give us the same note twice.
Feedback outranks your own earlier answers · a correction is permanent, not per-post
Step 06, in full
Most agencies will tell you they have a quality process. This is ours, whole — every check that runs on your article, named. Seventeen of them are pure arithmetic on the text, which is why they can run on every post instead of a sample.
Step 07, in full
Nine dimensions, ten points each, judged against your practice profile rather than a generic standard. An article can pass all nineteen mechanical checks and still be flat, subtly unsafe, or sound like nobody in particular. This is the part that catches that.
An example readout. Safety and scope are at the ceiling because they are gated; the rest is where the editing happens.
Is every claim true, hedged to what the evidence supports, and safe for a reader whose state you don’t know?
Inside your licensed scope, your age and format limits, and your jurisdiction’s advertising rules.
Non-shaming throughout, and never implying the reader’s difficulty is a failure of effort.
Would a reader be better off for having read it, or did it only describe the problem back to them?
Scored against your own writing sample. Not “does it sound warm” — does it sound like you.
Experience, expertise and authority shown through the writing rather than asserted in a bio line.
The hollow opener, the empty universal, the paragraph any practice could have published.
Findable for the searches that actually lead to a booked first session in your area.
Does it give a reader who is nearly ready a clear, low-pressure next step?
The hard gate
A low score on safety or scope blocks publication outright, no matter how strong the other seven dimensions are. A beautifully written article that drifts outside your licensed scope is not a good article with one flaw. It’s the one that costs you something.
AI visibility — AEO & GEO
When someone asks an AI assistant about therapy, it assembles an answer from a small number of pages it can read, trust, and quote cleanly. Being quotable is a property of how a page is written and structured — which means it’s something we can build in deliberately, and audit afterwards.
Each assistant selects sources differently — one favors recency, another structure, another consensus across sites. Every article is scored for AI citation readiness across the surfaces your future clients actually use:
This is the same discipline as SEO, not a replacement for it — the traditional signals still decide whether you rank, and ranking still feeds the assistants. We do both on every post, because doing one well and ignoring the other is how a practice ends up invisible in the room that’s growing.
Non-negotiable
These aren’t preferences we try to honor. They’re constraints written into the system, restated in every brief, and checked on every finished article. If a topic genuinely needs a clinician’s judgment call, the post stops and flags itself rather than guessing.
What you’re actually buying
A freelance writer gives you a draft you now have to check. A cheap AI tool gives you something you’d be embarrassed to publish, and the checking is still yours. What’s expensive here isn’t the writing — it’s everything built around it so that the article arriving in your dashboard is one you can read once and publish.
Ads stop the day you stop paying. An article that ranks keeps bringing people in for years, and the library compounds — each post makes the next one easier to rank and the practice as a whole more credible to both Google and the assistants.
A therapist writing one good, well-sourced, properly optimized post themselves is losing most of a day. Not a day of admin — a day of clinical hours, and it’s the first thing to be dropped when the caseload is full. That’s why practice blogs die in month three.
Scope, advertising rules, protected information, crisis language. Most content services have never heard of these constraints. Ours were built from them first, and they’re checked mechanically before anyone’s professional judgment is spent on the draft.
Your practice profile, voice profile, article archive and every correction you’ve given belong to you. So does everything we publish. If you ever leave, it all leaves with you.
That is where your voice profile starts. We’ll walk you through the intake, show you a post built the way yours would be, and tell you honestly if your practice isn’t ready for this yet.
Drafts land in your dashboard · nothing publishes without your yes