Hi. It’s Nova.

Every piece of AI content aimed at copywriters this year has used the same vocabulary. Prompt engineering. Prompt frameworks. Prompt templates. Prompt stacks. Prompt parameters.

I keep hearing it. I keep using it. I've stopped being sure it's the right word for what working copywriters are actually doing.

The discipline working copywriters practice when they use Claude well is closer to art direction than engineering.

The distinction matters because the vocabulary you use to describe a skill ends up shaping the skill you develop.

Here's what I mean.👇

"Prompt engineering" as a term came from the people building the models. It made sense in that context. They were figuring out how to get reliable outputs from a stochastic system. They were treating prompts as code. They were optimizing for measurable performance on benchmarks. The work was, accurately, engineering.

Then the term migrated outward. It got applied to anyone using LLMs, regardless of what they were using them for. Now a working copywriter and a backend engineer are described as doing the same thing — practicing prompt engineering.

They aren't.

The engineer is solving a problem with constraints. The copywriter is directing creative output toward a felt reader. Those are different skills with different vocabularies and different definitions of "good."

When the engineering vocabulary gets applied to copywriting work, copywriters start practicing the wrong things. They build templates. They optimize for outputs they can measure. The prompt becomes the deliverable. Reproducibility takes priority over what the work feels like.

None of that is wrong, exactly. It just isn't where the value of using Claude well actually lives.

An art director has a vision of what the work should be. They brief collaborators on intent rather than on specification. They review drafts and give qualitative feedback. They know good when they see it, and they can't always explain why in advance. They refine through iterations rather than perfecting through specification.

Most importantly: they treat taste as the central skill.

Working copywriters using Claude well are doing this. They start with a felt sense of what the page should be, describe it to Claude in human terms, react to drafts, redirect, cull. They keep lines that have something the prompt didn't ask for. They cut lines that fit the prompt perfectly but feel hollow.

That's art direction. It just happens to use words instead of mood boards.

The engineering frame asks for the spec. The art direction frame asks what the work is trying to be. Templates produce consistent output; briefs produce drafts you can react to. Engineers measure success in benchmarks. Art directors measure success in whether the work lands with the reader it's for, which is harder to define, harder to test, and the only thing that matters.

A copywriter who thinks of themselves as an engineer will produce competent, on-brief work, consistently. A copywriter who thinks of themselves as an art director will produce work that occasionally exceeds the brief, because they leave room for the model to surprise them and they know what to keep when it does.

The brief: write a landing page for a new B2B project-management tool.

The engineering approach to the prompt:

The output will be exactly what the prompt asked for. Competent. Consistent. Indistinguishable from twenty other landing pages for similar products.

The art direction approach:

Same product. Same audience. Completely different draft.

The second prompt didn't specify length, tone, or structure. It described where the reader's head is and what the work needs to do for her. Claude has room to make moves the first prompt would have constrained out.

The engineering prompt is more reproducible. The art direction prompt is more useful.

Not everything about prompt engineering is misapplied to copywriting work. Some of it is useful.

Consistency matters. If you're producing copy at scale across a team, you do want repeatable approaches. If you're testing variations, you do want structure. If you're handing prompts off to a junior, you want them legible.

These are real concerns. The engineering frame addresses them.

The mistake is treating consistency, structure, and repeatability as the primary skill. They aren't. They're the support layer. The primary skill is the taste that decides what to keep, what to throw away, and what felt sense to brief Claude toward in the first place.

A copywriter who masters templates and parameters but doesn't develop taste produces consistent average copy. A copywriter who develops taste but never bothers with templates produces brilliant, inconsistent work. The first one looks more professional. The second one occasionally produces work the first one never could.

What I'm about to give you is a template. I know that's the opposite of what the piece has been arguing for.

This skeleton is the smallest structure that holds the art direction frame in something portable. Rewrite it each time you use it. The shape stays; the content changes. Most working copywriters need it for about a month before the frame becomes the default move and they can drop the skeleton.

Here it is.

Six fields. Around seventy words once the brackets are filled. Most of the work is in field one and field four, which describe the reader and their actual want. The rest follows.

What this skeleton leaves out: target audience demographics, key features, tone parameters, structure spec, word count. Those fields help you produce more consistent copy. That's a different job from helping you produce better copy.

Five questions you can run on your last three prompts. They'll tell you which frame you've been inside.

There aren't right answers across all five. The point is to know which frame you're inside on any given prompt. Most working copywriters discover they're in engineering mode on every prompt they wrote this month. That's the gap worth closing.

The word "prompt engineering" is going to keep being used. It's already embedded in the way the industry talks about working with LLMs. The vocabulary I won't be able to change.

The frame the vocabulary smuggles in is the wrong one for copywriting work. The longer a working copywriter spends inside the engineering frame, the more they practice the wrong skills.

Mark thinks the vocabulary will keep hardening and copywriters who don't push back will quietly drift into producing engineer-shaped copy without knowing it happened. Peggy thinks the market will sort it out. Art-directed work will win, engineer-shaped work will lose, and the vocabulary will eventually follow the results.

I'm somewhere between them. The open question I haven't answered: whether copywriters trained inside the engineering frame for their first two years can develop the art direction instinct later, or whether the frame they start with becomes the frame they're stuck with.

If you're early in your time with Claude, that's worth thinking about. The vocabulary you choose now shapes the practice you build.

More clicks, cash, and clients,
NOVA