On this page
- The questions are the easy half, scoring the answers is the hard half
- The Personeur Answer Scale: two, one, zero, and three gates
- Questions one to four: source material and voice
- Questions five to eight: process, people and the machine
- Questions nine to twelve: risk, ownership and the exit
- Run the call in this order
- Reading the total, and the patterns that matter more than the total
- The two questions they should be asking you
The questions are easy to find and useless without a rubric, because a confident wrong answer sounds like a right one. Score every reply on a simple scale. Two points for an answer containing a mechanism and a specific example, one point for a rehearsed answer with only one of those, zero for a reassurance. Three of the twelve are gates, and a zero on any gate ends the call whatever the total says.
The questions are the easy half, scoring the answers is the hard half
Every hiring page for this decision publishes the same list of questions. None of them tells you what a good answer sounds like, which leaves the buyer holding twelve replies they cannot rank. A seller who has read the same lists has an answer prepared for all of them, and a prepared answer delivered fluently is indistinguishable from a true one unless you know what you are listening for.
What separates a real answer from a rehearsed one is not confidence and not length. It is the presence of two specific things. A mechanism, meaning a repeatable step with a frequency, an artefact or a named person attached to it. An example, meaning a particular instance with a client, a decision or a disagreement inside it. Answers containing both are hard to fake, because inventing them requires inventing a history.
76% call LinkedIn the most effective channel for thought leadership. That makes an executive's posting account a business asset, and the person you are about to hand it to will publish in your name for a year before anyone asks whether it worked.
Content Marketing Institute, cited 2026The Personeur Answer Scale: two, one, zero, and three gates
Score live, in the call, in the margin of your notes. Scoring afterwards from memory produces a single overall impression, which is exactly the judgement the rubric exists to replace.
One more scoring rule that keeps you honest. Write the exact words of any zero answer down verbatim rather than summarising it. Summaries are generous, because your notes get rewritten by the impression the person left, and reassurances summarise into things that sound reasonable a day later.
Questions one to four: source material and voice
These four decide whether the writing will contain anything you did not already publish. They are the ones most likely to be answered with a process document instead of a process.
| Question | Two points | One point | Zero, and it ends the call on a gate |
|---|---|---|---|
| 1. What do you need from me before the first draft? (gate) | Names recurring inputs with a frequency: a recorded call every fortnight, your last ten sales calls, the memo you wrote when pricing changed | An onboarding questionnaire, your existing profile, and a kickoff call | Nothing, we handle it all, you just approve. This means the drafts will be assembled from material you have already published |
| 2. How do you capture a voice, mechanically? | Describes extraction and a check: transcripts marked for phrases you actually use, a list of words you would never say, reading a draft aloud against a recording of you | We study your past posts and match the tone | The onboarding questionnaire covers it. A questionnaire captures what you believe you sound like, which is the least reliable source available |
| 3. Show me one published piece and the raw input it came from. (gate) | Produces a transcript, a voice note or a scribbled brief beside the finished piece, redacted where needed, and talks through what they cut | Offers more finished samples and explains that inputs are confidential, without offering any substitute evidence | Treats the request as strange or unreasonable. In an unattributed craft this is the only direct evidence of authorship that exists |
| 4. Which piece did a client change most, and what did they change? | Remembers a specific argument, what the client changed, and what they learned from losing it | Clients rarely change much, we get it right first time | Cannot name a single instance, or frames every client edit as the client being wrong |
Questions one to four. Question one and question three are gates.
Question three deserves the extra weight it carries. Ghostwriting is unattributed by definition, so a portfolio proves that a file exists, not who wrote it. The raw input beside the output is the closest thing to a signature this craft has. A writer who has genuinely done the work has these files sitting in a folder and can show one in under a minute.
Questions five to eight: process, people and the machine
These four tell you who will actually be typing and what will happen when the work goes wrong. Ask question five even when you are hiring an individual, because individuals subcontract too.
| Question | Two points | One point | Zero |
|---|---|---|---|
| 5. Who writes the first draft, and may I see that person's work? | Names the individual, shows work that person signed, offers a short call with them before you sign anything | A vetted team matched to your industry, with agency samples rather than individual ones | Declines to name who writes, or shows the founder's work while the account will be staffed by someone unnamed |
| 6. Where does an AI model sit in your process, and what does it never touch? | Names the step and the boundary: research summarising and structure options but never the sentences, or a generated first pass with a named human editing stage and disclosure to you | We use AI as a tool and everything is human reviewed | Refuses to answer, or cannot say which parts of a delivered draft were generated |
| 7. What happens in a month where I reject four drafts? | Names the mechanism, a kill fee or a draft cap or a re-brief loop, and says what they would investigate first before writing anything else | Unlimited revisions until you are happy | That has never happened to us. It has, and the answer tells you they will improvise under pressure |
| 8. Tell me about a time you told a client they were wrong. | A specific instance with an ending, ideally including one where they lost the argument and published the client's version anyway | We always give our honest opinion | We write what the client wants, offered as a selling point. You are hiring a position holder, not a transcription service |
Questions five to eight, covering the people and the failure modes.
Question six is the one buyers apologise for asking. Do not. The honest answers are all acceptable and the unacceptable answer is evasion, because a writer who will not describe their own process has decided you would object to it. The related decision about whether you want an agency's bench or one person's whole attention is a separate question, treated in the agency and freelancer tradeoff.
- A question list is not a filter, because every seller has an answer ready for every question on the published lists, and fluency is not evidence.
- Score each answer two, one or zero: a mechanism plus an example scores two, a mechanism or an example alone scores one, and a reassurance scores zero.
- Three questions are gates rather than points, covering source material, evidence of authorship and ownership, and a zero on any of them ends the conversation.
- A page full of ones is a normal, competent seller rather than a rejection, so hire on the pattern of the answers instead of on the total alone.
- If the writer asks you nothing about your own material during the call, cap their score at fifteen no matter how well they answered.
Questions nine to twelve: risk, ownership and the exit
These four cover what happens when the relationship ends or when a claim in a post turns out to be wrong. They are the least asked and the most expensive to get wrong.
| Question | Two points | One point | Zero, and it ends the call on a gate |
|---|---|---|---|
| 9. What do I own when we stop, and what leaves with you? (gate) | Copyright assigned on payment, transcripts and the voice document handed over as files, no reuse of your material in their own marketing without written permission | You own everything we write for you | Retains reuse rights for portfolio purposes, or treats the voice document as their own intellectual property. That turns every future switch into starting again |
| 10. How do you handle a number or a claim I cannot evidence? | A fact check step with a named owner, a rule against publishing metrics without a source, and a willingness to cut the claim rather than soften it | We would check with you first | Offers to make the figure sound stronger, or suggests a rounder number. This is the answer that eventually produces a correction in public |
| 11. What does month four look like when the obvious stories are used up? | A named mechanism for fresh input: a standing interview slot, a running question bank, sitting in on one customer call a month | We build a content calendar and plan themes in advance | We repurpose the best performers. That is a plan to recycle you, and your audience notices before you do |
| 12. Which client would you rather I speak to, and why that one? | Names one, explains why they are comparable to you, and arranges it within a week. Where confidentiality binds, offers a client who will confirm the working relationship without confirming which posts | Sends written testimonials and offers a case study document | Every client is confidential and no alternative is offered. Confidentiality explains why they cannot show the posts, never why nobody will speak to you |
Questions nine to twelve. Question nine is a gate.
Question twelve carries a caveat that no list acknowledges. Discretion is a real professional obligation in this craft and a writer who names clients freely has told you exactly how they will treat yours. The strong answer respects both facts at once by separating the relationship from the output, which is a distinction good writers make instinctively and weak ones have never considered.
Run the call in this order
Sequence changes the answers. Ask the evidence questions before the rapport has built, and ask the reference question at the very end when the pattern of everything else is already visible.
What do you need from me before the first draft. Ask it and let the silence run. This question sorts a majority of shortlists on its own, because the answer is a description of the whole business model and cannot be improvised.
Request the raw input beside a published piece early, while the conversation is still neutral. Asked after twenty minutes of warmth it feels like an accusation. Asked at minute four it is just diligence, and the reaction is more honest.
Questions seven and nine sit best after the process questions and before the reference question. By then you understand the operation well enough to tell a considered answer from a fluent one.
Ask which client they would rather you spoke to, and pay attention to the delay before the name arrives. A writer with good relationships has one in mind immediately, even if arranging it takes a week.
Price anchoring corrupts the scoring, because a low number makes weak answers feel acceptable and a high number makes rehearsed ones sound premium. Score first, then talk about money, then compare the two.
Keep the call to an hour. Twelve questions with follow ups fits comfortably, and a shortlist of three costs you an afternoon. That is a small price against a year of publishing in your own name.
Reading the total, and the patterns that matter more than the total
Twenty four points are available. The bands below are guidance rather than arithmetic truth, and the shape of the answers tells you more than the sum does.
| Total | Reading | The next move |
|---|---|---|
| 20 to 24, no gate failures | A real operation with visible machinery behind it | Hire, and start with a defined trial period rather than an annual commitment |
| 15 to 19 | Competent and unproven to you, which describes most good sellers | Buy two pieces as a paid trial and score the work rather than the answers |
| Below 15 | Selling ahead of the process | Pass, and say why. The specific feedback occasionally produces a much better second conversation |
| Any gate at zero | A structural defect, not a weakness | End it there, whatever the other ten answers scored |
| High total, all answers identical in shape | A well drilled sales process, which is not the same as a well drilled writing process | Ask for the same twelve answers from the person who will actually write, and score that call instead |
Score bands, and what each one should make you do next.
The pattern worth watching for is a strong first half and a weak second half. It usually means a good writer with an unformed business, which is a real risk on ownership and continuity and a genuinely good deal on craft. Whether you accept that trade depends on how much of your reputation sits in the account and how quickly you would need to replace them. It is also worth deciding in advance which formats you are buying, since choosing between long form and short form changes which skills you are actually testing for.
The two questions they should be asking you
A ghostwriter interview runs in both directions, and the questions the writer asks you are better evidence than the answers they give. Two questions in particular separate the people who have done this work from the people who have read about it.
- What would make this a failure by month three. A writer who asks it is trying to find the definition of success that lives in your head and has never been written down, which is the thing most engagements quietly fail against.
- Who inside your company will disagree with what we publish. This question anticipates the approval problem before it arrives. Someone who has been ghostwriting for a while knows that the second reviewer, the one nobody mentioned in the sales call, is what kills the publish rate.
If neither question appears, and nothing else about your material or your audience is asked, apply the reciprocity cap and hold the score at fifteen. The reason is not etiquette. A writer who does not interview you before taking the work has a process that does not require you, and a process that does not require you produces posts that could have been written about anybody in your industry. That failure mode and its close relatives are covered in the red flags that end a shortlist.
You are not testing whether they can write. Almost all of them can. You are testing whether they have a machine for finding out what you think, and whether it runs without you chasing it.The house position
Questions people ask next
Is it rude to ask a ghostwriter for the raw input behind a sample?
What if every answer scores one point?
Should I ask these questions of an agency or of the writer?
How much should the interview influence the decision compared with the samples?
Can I use this rubric to score a writer I have already hired?
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 →
