On this page
- The flags everyone lists are the cheap ones to catch
- Drift one: the person who sold you is not the person writing
- Drift two: the drafts stop containing anything you said out loud
- Drift three: your hooks start appearing under other people's names
- Drift four: the strategy call quietly became an approval call
- The Personeur Drift Audit, run at the end of month two
- What to say when you find one, and what a real answer sounds like
- Three pre-sale flags that still survive contact with reality
- When the drift is yours and not theirs
Pre-sale warning signs are the cheap ones to catch. The expensive failures start around week six, when the writer who sold you quietly becomes someone junior, your drafts stop containing sentences you actually said, your hooks start appearing under other names, and the strategy call becomes an approval call. All four are measurable against your own month one baseline, and all four are fixable if you name them early.
The flags everyone lists are the cheap ones to catch
A missing portfolio, vague pricing and a guaranteed follower number are genuine warning signs, and every page written on this subject lists them. They are also the easiest problems you will ever have, because you find them before any money moves. What nobody has written down is the set that costs you a quarter.
The expensive failures are quiet. They do not announce themselves in a proposal. They arrive in week six or week nine, inside an engagement that still looks healthy on the invoice, and they share one property: the writing keeps arriving on time while the thing you were actually buying quietly stops being delivered.
of B2B marketers call LinkedIn the most effective channel for thought leadership. That is why an executive feed is worth paying a professional to produce, and it is why a quietly degraded one is worth catching in month two rather than month eight.
Content Marketing Institute, cited 2026The four drifts below are diagnosable. Each one has a signal you can measure, a baseline you already own from your first month, and a question that produces a real answer when you ask it out loud.
The first month of any ghostwriting engagement is unrepresentative. The writer is still interviewing you, the senior person who sold the work is still attending calls, and everyone is on their best behaviour. Month two is the first honest month. Month three is when whatever month two established becomes the permanent shape of the work.
Drift one: the person who sold you is not the person writing
A writer swap is not misconduct, and it is the most common single cause of a quality drop in month two. Studios and agencies sell with their strongest voice and staff with their available one. The swap is rarely announced, because nothing in the agreement required an announcement.
The tell is not in the prose, which is why reading the drafts more carefully will not find it. The tell is in the questions. A writer who ran your interview knows what you said about the product launch that failed, and will reach for it without being reminded. A writer handed a transcript uses the parts that transcribe well and skips the parts that needed the room.
The test that takes ninety seconds
Reply to a draft with a question that can only be answered from inside the interview rather than from the transcript. Something close to: you asked me about the pricing argument with my co-founder, do you want that in this one or should we hold it. A writer who was in the room answers in a line. A writer who was not either goes quiet, asks you to remind them, or answers a slightly different question.
None of this proves bad faith. It tells you who is holding the account, which is a thing you are entitled to know and were probably never told. The right response is a name and a role, not a reassurance about process.
Drift two: the drafts stop containing anything you said out loud
Voice drift arrives as drafts that are competent, publishable and completely unattributable to you. The mechanism is input starvation. Interviews get shorter, then get replaced by a shared document, then by nothing at all, and the writer fills the gap with the general knowledge of your industry, which is by definition the thing that sounds like everybody.
It is measurable without any benchmark. Take your last four drafts and highlight every sentence you can trace to something you genuinely said, wrote or sent: an interview recording, a voice note, a message, a document you handed over. Divide the highlighted sentences by the total. That number is your source-trace rate.
Now do the same for four drafts from your first month. The absolute figure does not matter and no published benchmark for it exists. The comparison is the finding. As a working rule, a fall of more than a third against your own month one figure means the input has stopped and the writer has moved from transcribing you to generating around you.
Voice drift also hides more easily in a 900 character post than in a 1,200 word piece, which is one reason the choice between a newsletter and a run of posts changes what you are able to detect from the outside.
| What changed upstream | How it looks in the draft | The fix that actually works |
|---|---|---|
| Interviews got shorter | Correct claims, no specifics, no numbers you recognise | Restore one recorded thirty minute interview per fortnight rather than per month |
| Interviews became a written brief | Tidy structure, flat middle, none of your digressions | Send voice notes instead of documents, because speech carries syntax that writing removes |
| Input became your own published posts | The engagement is recycling you and gently narrowing | Hand over raw material nobody has seen: internal memos, customer emails, meeting notes, half-finished arguments |
| Input did not change and drift happened anyway | Vocabulary that is not yours, positions you would not defend | Ask who is drafting, because this is drift one wearing a different coat |
What a falling source-trace rate usually means, and the change that fixes it.
Drift three: your hooks start appearing under other people's names
Template drift is the only one of the four you can verify entirely from outside the engagement, and it is the one buyers almost never check. A writer under volume pressure converges on the structures that performed, then runs those structures across the roster. Your opening line becomes a shape that three other founders are also using this month.
Around 3% of members post more than once a week, according to aggregate 2026 LinkedIn statistics reports. A writer serving many accounts is therefore producing a meaningful share of the consistent output in any narrow niche, which is exactly the position in which reusing what works stops feeling like a compromise and starts feeling like craft.
Take the first line of your last five posts and write down its structure rather than its words. A hook that runs most founders think X, they are wrong, and here is the number that proves it is a shape, not a sentence.
Search the distinctive middle clause in quotation marks, then search the shape in your own words. You are looking for the same move performed with different nouns, which is what convergence looks like from outside.
Most writers name two or three clients somewhere. Read four consecutive posts from each. If the opening move, the paragraph rhythm and the closing question are the same across two different people, they are the same across you.
Do you run structures across accounts, and how do you decide when to stop. There is a good answer to this and it involves a stated rule. The absence of any rule is the finding.
Convergence is not theft and it is not always wrong, because a structure that works is a structure that works. It becomes a problem at the point where your posts read as a genre rather than as a person, which is precisely the point at which the executive who hired the writer stops being the reason anyone reads. If you want the reverse of this test, applied before you hire rather than after, it is the same instinct used in reading samples the way an editor does.
- The warning signs that cost real money are invisible before you sign, because they are properties of how the work gets produced rather than how it gets sold.
- Voice drift is measurable without any external benchmark: highlight the sentences in a draft you can trace to something you actually said, and compare the ratio against your own first month.
- A writer swap is not misconduct, but an unannounced one tells you that nothing in your agreement required anyone to tell you who is holding the account.
- Template drift is the one flag with no innocent explanation, because running a hook structure across a roster is a production decision rather than an accident.
- Two of the four drifts are usually caused by the buyer, so count your own recorded input before you change supplier and reproduce the same result elsewhere.
Drift four: the strategy call quietly became an approval call
Agenda drift is the most expensive of the four, because it removes the part of the engagement that justified the price. You bought judgment about what to say. Somewhere around week seven the standing call stopped opening with what should we argue this month and started opening with here are four drafts, any changes.
The mechanism is not laziness on either side. Approval is a meeting that always has an agenda, because drafts exist and drafts create their own business. Strategy is a meeting that requires preparation from both parties. When either party has a difficult week, the call defaults to the version that fills itself.
Measure it in minutes. On your next standing call, note the minute at which the first draft appears on screen. If a draft is up inside the first five minutes of a thirty minute call, and it happens twice in a row, you are running an approval meeting on a strategy invoice.
The fix is structural rather than personal. Split the two things. Approval belongs in writing with a deadline attached to each stage, and if this is the drift you have, the repair is a real approval workflow with stage deadlines rather than a firmer conversation. Strategy belongs on the call, opened with a question neither of you has answered yet.
The Personeur Drift Audit, run at the end of month two
Run the audit at the end of month two and again at the end of month five. It takes about forty minutes and it produces four readings you compare against your own baseline rather than against anybody's published benchmark, because no trustworthy benchmark for any of these exists.
Three of the four have honest explanations that are nobody's fault, and the audit is built to start a conversation rather than deliver a verdict. The exception is the shape check, because running structures across a roster is a production decision made by an adult, and the only useful answer to it is a stated rule about where the reuse stops.
What to say when you find one, and what a real answer sounds like
Name the observation and never the motive. Each of these has a legitimate explanation, and the difference between a supplier worth keeping and one worth leaving is entirely visible in how the question is answered rather than in whether the problem existed at all.
| What you found | What to ask | An answer worth keeping | An answer worth leaving over |
|---|---|---|---|
| The drafts changed hands | Who is drafting now, and who runs the interviews | A named person, their role, and an offer to put them on the next interview | A reassurance about process, a reference to the team, no name given |
| Source-trace rate fell by half | How many recorded minutes did you get this month against month one | An honest count and a request for more time with you | A claim that the writer has internalised your voice and no longer needs input |
| Two hooks matched other clients | Do you run structures across accounts, and where do you stop | Yes, here is the rule we use and here is the line we do not cross | Denial, or an explanation that these are simply industry conventions |
| The strategy call is an approval call | Can approval move to writing so the call holds the argument | Immediate agreement plus a written agenda for the next one | Agreement in principle followed by an identical call two weeks later |
Four findings, the question that opens them, and how to read the reply.
Ask one at a time and in that order. Four grievances delivered together read as a prelude to cancellation, and the supplier will answer the mood rather than the question. One observation, answered properly, tells you whether the next three are worth raising.
Three pre-sale flags that still survive contact with reality
The standard pre-sale lists are not wrong, they are incomplete, and three items on them are worth keeping. The rest are preferences dressed as rules.
- A guarantee expressed in something the writer does not control. Impressions, follower counts and inbound replies are decisions made by the platform and by readers. A guarantee attached to any of them is either a misunderstanding of the mechanism or a sales device, and both are disqualifying.
- A process that opens with drafting. If the first deliverable is a post rather than a short set of positions you have agreed to defend in public, the engagement has no source material, and month two voice drift is already on the calendar.
- No named point of contact for the writing itself. A salesperson, an account manager and an unnamed writing team is a structure that makes the first drift invisible by design, which is usually why it is structured that way.
A writer with no public portfolio may simply be discreet, and in this field discretion is a qualification rather than a gap. Vague pricing is often a scoping problem rather than an evasion. Before assuming the operating model is the fault, it is worth understanding what an agency and a solo writer each risk, because the two structures fail in opposite directions.
When the drift is yours and not theirs
Two of the four drifts are frequently caused by the buyer, and an engagement starved of input will fail regardless of who is writing. If you cancelled three interviews this quarter, the source-trace rate fell because you stopped speaking, and no supplier can repair that from their side of the table.
A writer can only be as specific as the material they receive. If the last original thing you handed over arrived in month one, the drafts have been assembled from your published archive, and the narrowing you are noticing is your own voice being fed back to you. The audit is still worth running, because it tells you which of the four is yours to fix.
The reverse test is a single number. Count the recorded minutes you gave the writer last month. If the figure is under sixty, fix that before changing supplier, because you will reproduce the same result with the next one and pay for a fresh onboarding to do it.
- Count the recorded interview minutes you actually supplied in the last month
- Count the pieces of unpublished raw material you handed over: memos, customer emails, meeting notes, arguments you abandoned
- Check whether you answered draft comments inside the agreed window or three days after it
- Check whether you killed ideas at the brief stage or at the draft stage, because killing at draft stage burns a week of somebody's month
- Ask whether you have changed what you want to be known for since the engagement began, and whether you told anyone that you had
Run those five, then run the audit. If the buyer side is clean and two or more thresholds have been crossed, you have an evidenced conversation rather than a feeling, and evidenced conversations tend to be answered rather than managed.
Questions people ask next
Is it normal for a ghostwriting agency to change my writer without telling me?
How do I tell voice drift apart from my own writing simply improving?
What should I do if I find my hook structure on another founder's feed?
How long should I wait before raising a problem with a ghostwriter?
Can a ghostwriter fix voice drift without more of my time?
Send one old post.
Email any post you have published, with your LinkedIn URL. A rewrite in your voice comes back free, so you can judge the craft on your own words.
Get the free rewrite →
