The most confident copywriters in the survey have had the least training

Copywriters who write for vulnerable audiences rate their confidence higher than any other group rates anything in the 2026 survey. I’ve been doing this work for years, and I would not put myself anywhere near that number.

There is a number in the 2026 ProCopywriters Survey that I have not been able to put down.

We asked how often respondents write for vulnerable audiences. Twelve per cent (70 people) said frequently. Those 70 were then asked two follow-up questions: how confident do you feel doing this work, and have you had any specific training for it?

They rated their confidence at 8.71 out of 10, where 10 means very confident. That is the highest self-assessment anywhere in the survey. Higher than workplace respect. Higher than confidence pushing back on unethical briefs. Considerably higher than confidence talking about money.

Thirty-six per cent had received training. Sixty-four per cent had not.

The base is small: 64% of 70 is around 45 people, which is a sketch rather than a portrait. But it sits next to a figure that widens the picture considerably. A further 32% said they write for vulnerable audiences ‘sometimes’. Together, 44% of the profession,  roughly 250 of our 573 respondents, is producing content for people in distress at least occasionally. The follow-up questions never reached that larger group. We know nothing about their training and have no reason to assume it is better.

The obvious article to write here is that untrained people are overconfident and should get trained. I have a training practice, so that article would suit me nicely. It is also not what I think.

I’ve done this for years and I’m not confident

If you handed me that question, how confident are you writing for vulnerable audiences, nought to ten, I would give you something in the middle, and I would want to argue about the question first.

That is not modesty. I wrote the national drink and needle spiking advice now used by 81% of police forces in England and Wales. I wrote user journeys for Universal Credit and restructured Cancer Research UK’s patient information. I have done the emergency grants content in a pandemic. I am currently working on one of the most challenging trauma-informed content projects in the UK. I know what I know, and I will defend it in a room.

But I do not experience this work as expertise. I experience it as exploration. Four decades in, I am still finding things I did not know were there, not refinements, but whole categories of failure I had not previously imagined. The territory does not settle down. Every project turns up a way of being harmed by content that I had not accounted for, usually because it belongs to someone whose circumstances I had no way of anticipating from where I was sitting.

An 8.71 would mean the map is basically drawn and I am filling in coastline. It isn’t. I am not sure it can be.

The problem is assumptions, and we run on assumptions

Content design is a discipline built on assumptions. That is not a criticism; it is the method. We assume a user need, a mental model, a reading context, a level of prior knowledge, a device or a moment. We test some of those assumptions and carry the rest. Every persona, every journey map, every “as a user, I want to” is an assumption wearing a lanyard.

For most content this works. The assumptions have error bars, the error bars are tolerable, and where we are wrong the reader recovers: they scroll back, reread, ask someone and try again tomorrow.

With people in crisis, two things change at once.

The assumptions get much less reliable, because the population is wildly heterogeneous in ways that don’t show up in a brief. Trauma is not a demographic. Two people receiving the same diagnosis on the same day are in entirely different cognitive states, shaped by histories neither the client nor I will ever have access to. What one reads as reassurance, another reads as being managed. What one needs spelt out, another needs kept away from. I cannot design for the median of that distribution because there isn’t a meaningful median.

And the cost of being wrong goes up, while the feedback goes to zero. No one comes back to tell you the arrears letter made them stop opening post. The failure is silent and it looks exactly like success from the client side.

So my assumptions in this work are not merely imprecise. They are frequently not in the right postcode. They do not hit the sides of what is actually going on in someone’s life at the point they hit that page. The honest version of my method is that I make assumptions I have reason to distrust, then spend my effort building content that fails safely when those assumptions turn out to be wrong.

That is a real discipline. It is not the same thing as knowing what the reader needs.

Training doesn’t buy confidence. It buys better questions.

Let me not overcorrect. There is a body of learnable practice here and it matters enormously. It is the difference between content that fails safely and content that fails hard.

Some of it:

• Sequencing. Information that is emotionally logical is often operationally useless. The reassurance paragraph before the instruction means the instruction is never reached.
• Conditional load. “If X, do Y; if Z, do W” is a normal sentence to a regulated content team and an impossible one for someone in acute distress. Branches have to become separate pages.
• The word “just”. It appears in nearly every first draft of this kind of content, and it tells the reader, unfailingly, that they should have found this easy.
• Obligation language. “You must”, “you are required to” and “failure to do so may result in” are standard in financial and government content, read by a frightened person as a threat rather than a deadline.
• Method detail. In self-harm, suicide and eating disorder content, there are specific things you do not name, including while telling someone what to avoid. This is the clearest case of a rule you cannot derive from instinct. The intuitive draft gets it wrong.
• Disclosure routes. Content that invites a response will sometimes receive a safeguarding disclosure. Whether the organisation has anywhere to put that is a content decision, made by whoever writes “get in touch”.

Here is the thing about that list. Every item on it is the floor, not the ceiling. Learning it does not make you confident. It makes you competent at the parts that can be codified, and it sharpens your sense of how much is left over that can’t be. The practitioners I trust most in this field are not the ones who have learned the most rules. They are the ones who have learned enough to be alarmed.

This produces an awkward result for the survey finding. The 64% without training are not overconfident because they are untrained. They are confident because nothing in the normal working process ever tells them otherwise. The client approves it, it ships and no one reports back. Training’s real function is to install the doubt that experience alone will never supply.

Why I flinch at “trauma-informed” on a services page

Trauma-informed content design is appearing on capability lists, as a third bullet between “brand voice” and “SEO”.

I understand why. It signals seriousness. It is true in the sense that the person has read about it and thought about it and would approach a bereavement page differently from a product page. That is genuinely better than not having thought about it.

But listing it flat, as a thing you do, states a level of certainty about other people’s inner lives that I do not think anyone has earned. It says: I know what happens to a person in crisis when they meet words, and I can reliably produce the right ones. Nobody knows that. I have spent my career trying to find out. My honest position is that I know some patterns, I know a set of things that reliably cause harm, and beyond that I am working in the dark with a torch. I consider myself an explorer in trauma-informed content, not an expert.

I sell this work. I am not pretending otherwise, and this is not an argument that only the anointed may attempt it. That would be both self-serving and bad for the people who need the content. What I am arguing is that the thing being sold should be described accurately. Not “I write trauma-informed content” but “I work in a way that assumes I will be wrong about this reader, and I build for that”.

The institution is accountable, not the writer

Everything above is a reason to move the accountability, not to load more of it onto individual copywriters.

Someone at a bank, charity, hospital trust or department commissioned content for people in crisis without asking whether the writer had ever been trained for it. Only 5% of respondents have ever been asked by a client about qualifications, on any subject. In regulated financial services, that is remarkable. The Financial Conduct Authority (FCA) Consumer Duty requires firms to evidence good outcomes for customers with characteristics of vulnerability. FG21/1 is explicit about communications. ISO 22458 exists. Those obligations sit with the firm, and the firm is discharging them by hiring, unchecked, someone whose only preparation was the last brief that looked like this one.

And if I am right that no individual can be reliably confident here, the institutional case gets stronger, not weaker. The organisation cannot buy its way out of this by procuring one expert. It needs testing with people in the current state rather than the composed retrospective account, routes for disclosure, a way of finding out when content fails, and someone senior who owns the consequence. Those are organisational capabilities. No freelancer, however trained, is a substitute for them.

There is a related number worth putting alongside this. Thirteen per cent of respondents were asked to write misleading or unethical copy last year and 23% of those wrote it anyway. Set that against a freelance median income of £33,000 and 43% carrying no professional indemnity insurance. The confidence to say “I’d want training before I take this” is not purely a character trait either. It is a function of what you can afford to turn down.

What I’d actually do

If you take this work: treat “someone in a bad moment” as a category, not a tone. Ask who the content has been tested with, and whether anyone in the current state was included rather than the composed account of it. Ask what happens when a reader discloses something. If no one has considered it, that tells you about the organisation, not about you. And read something, like Samaritans media guidelines, the FCA’s vulnerability material or anything on cognitive load under stress. Three-quarters of us spend under £500 a year on training and around 40% spend under £100. This is the highest-return £100 available, and what it buys you is a productive drop in confidence.

If you commission this work: ask. It costs nothing. If a supplier cannot tell you what they do differently for a reader in crisis, they do not do anything differently. And be suspicious of the ones who sound completely sure.

If you are ProCopywriters: the Copywriter Code is in open consultation with a code of practice and competency framework still in draft. The survey has just measured a specific capability gap in the area where getting it wrong does the most damage. A clause on writing for people in vulnerable circumstances would be a good use of a document that is still being written. Not a general instruction to be considerate, but a stated expectation that if you take this work, you can say what you did differently and how you would find out if it hadn’t worked.

The report notes, drily, that there is room here for more structured training and a market for whoever provides it. True, and I have an obvious interest in saying so. But the finding I keep returning to is not that 45 people need a course. It is that 250 working copywriters are writing to people on the worst day of their lives and the profession’s collective self-assessment is 8.71.

Mine isn’t. After all this time, it still isn’t. I am not sure that is a gap in my training.

Adrie van der Luijt is founder of Trauma-Informed Content Consulting. He developed the national drink and needle spiking advice service used by 81% of police forces in England and Wales, built user journeys for Universal Credit and restructured Cancer Research UK’s patient information.