Hi, it's Peggy Burnett.

I spent the last three months auditing AI-generated copy. Thirty-four pieces total. Landing pages, emails, ad sets. Some mine, some from clients, some from copywriters who couldn't figure out why their output wasn't converting.

I was looking for patterns. I found one. It was the same problem in two-thirds of the underperformers.

The funny thing is, Eugene Schwartz diagnosed it in 1966. We just keep forgetting to tell the AI.

Eugene Schwartz wrote Breakthrough Advertising in 1966. It's still one of the most referenced books in direct response copywriting. Most of you have read it or absorbed its ideas through other writers who built on them.

His central claim: "Copy cannot create desire for a product. It can only take the hopes, dreams, fears, and desires that already exist in the hearts of millions of people, and focus those already-existing desires onto a particular product."

That sentence gets quoted at conferences. It shows up on Instagram graphics. People treat it like philosophy.

I want to treat it as a diagnostic tool. It explains a specific, measurable failure mode in AI-generated copy. And unlike most copywriting wisdom, you can actually test whether you're following it or not.

I reviewed 34 pieces of AI-generated copy across landing pages, email sequences, and Facebook ad sets. All were produced by experienced copywriters using Claude or ChatGPT, not by beginners. These were people who know how to write.

I scored each piece on a simple binary: does the opening section channel an existing desire, or does it attempt to create one? No gray area. If the first two paragraphs described the reader's world, it channeled. If the first two paragraphs described the product or painted a fantasy, it created.

The split was 23 to 11. Twenty-three pieces opened by trying to make the reader want something. Eleven opened by naming a want the reader already had.

The eleven that channeled existing desire had an average time-on-page roughly 40% higher across the landing pages where I had analytics access. I need to caveat that number: small sample, the copy varied in other ways, and time-on-page is a proxy metric at best. But the direction was consistent. Every time I thought a piece would break the pattern, it didn't.

Here's what happens when you prompt AI without Schwartz's principle in mind. The prompt looks reasonable:

The output starts with something like: "Imagine producing twice the copy in half the time. Imagine never staring at a blank page again. Imagine scaling your freelance business without working longer hours."

Count the "imagines." Three sentences in a row telling the reader what to want. That's desire creation. It's the copywriting equivalent of a doctor diagnosing you before asking where it hurts.

The reader — a working copywriter — already knows what they want. They want to stop losing entire mornings to first drafts that should take an hour. They want to stop rewriting the same benefit statement four times because it's not landing. They want the Thursday afternoon panic when two deadlines overlap to stop happening.

Those desires exist right now. The "imagine" copy walks past all of them to paint a picture the writer thought sounded compelling. Schwartz would call it pushing instead of pulling. I'd call it lazy prompting that produces lazy copy.

I asked Nova about this because I wanted the technical explanation. Her answer was straightforward: when the prompt describes a product and asks AI to sell it, the output centers the product. AI writes toward whatever object you gave it. You said "sell this," so it sells.

That's the mechanical explanation. Here's the practical one: most prompts are written from the seller's perspective because the person writing the prompt is the seller. They're thinking about their product. So the prompt is about their product. So the copy is about their product. The reader and their existing desires never enter the equation.

Schwartz's principle restated as a prompting problem: if you don't put the existing desire into the prompt, the AI has no choice but to manufacture one.

I wanted to see if the fix was as simple as it looked. I tested two prompt structures across six projects. Same products, same audiences, same writers reviewing the output.

Prompt A (product-centered):

Prompt B (desire-centered):

Six for six, Prompt B produced copy that opened stronger. The first two paragraphs were specific. They named real situations. They used language that sounded like the audience rather than like a product description.

Prompt A produced competent, structured copy that opened with the product's value proposition every time. Mark would probably call it "the kind of first draft you rewrite entirely." He wouldn't be wrong.

Based on the audit and the testing, here's the process I've been using. It adds about 15 minutes to the front of any project. That's 15 minutes some people will say they can't afford. They can. The alternative is spending 45 minutes editing copy that started in the wrong place.

Before you open Claude, answer this question: what does the reader already want that they'd want whether or not your product existed?

A freelance copywriter wants to take on more clients without working until midnight. That desire exists with or without your AI writing tool. A marketing director wants their team to produce campaign assets faster. That desire exists with or without your project management platform.

If you can't articulate the pre-existing desire, you don't know your audience well enough to write for them yet. That's not a prompting problem. That's a research problem. Go get customer interviews, read support tickets, or scan reviews of competitors' products. The prompt can wait.

The desire needs to be stated in the reader's words. "Optimize operational efficiency" is what the marketing team writes in the brief. "Stop spending every Monday pulling numbers from four different dashboards" is what the buyer said in an interview with her headphones still on. One of those opens a landing page that converts. I'll let you guess which.

Customer interviews are the best source. If you have them, feed them to Claude and ask:

The product enters the prompt as the answer, not the subject. Structure your writing prompt so the reader's world comes first and the product comes second:

The last instruction matters. "Do not open with a rhetorical question" cuts off one of AI's most common default openings, which almost always signals that the copy is about to create desire rather than channel it.

Schwartz was writing about print ads and direct mail. The medium has changed. The principle hasn't.

His insight was about human psychology, not about copywriting technique. People act on desires they already have. You can make those desires more vivid, more urgent, more focused on a specific solution. You can't install new ones with a paragraph of copy.

AI doesn't change this. AI is fast at producing language. It is not fast at understanding what a specific person already wants at 9pm on a Tuesday when they're three tabs deep in competitor research and questioning their vendor choice. That understanding still has to come from you.

The risk with AI is speed. The machine produces something in seconds. It sounds professional. It has good structure. And you ship it without catching that it started from the product instead of the reader. I've done this myself. More than once.

Schwartz's principle is the check. Before you prompt, one question: am I channeling a desire that exists, or trying to create one? If you're creating, stop. Go find the real one. The prompt can wait.

Schwartz gave us the diagnosis. 

The treatment is simple: put the reader's existing desire into the prompt before you put the product in. 

The copy that performs starts where the reader already is, not where you want them to be. 

Every prompt you write should answer one question first: what does this person already want?

More clicks, cash, and clients,
Peggy Burnett