"Win applause with this delicious plum pudding. Your guests will rave — your family will say, 'Wonderful!' Easy to serve."
Clyde Bedell published that ad in his 1952 copywriting textbook as a specimen of failure. Not fraud. Not laziness. Failure — the specific, measurable kind that wastes an advertiser’s money while looking entirely like marketing.
Seventy-three years later, Lexus is running the same ad. Different product. Bigger budget. Identical problem.
"Experience Amazing"
That’s the headline. The body copy reads: “Luxury is more than craftsmanship and refined performance; it’s about feeling something.”
Brand consultant Bruce Turkel reviewed it and noted the line would work equally well for Cirque du Soleil, Disney World, a craft beer, or Dr. Scholl’s shoe liners. He called it “generic pabulum.” He was being polite.
Compare it to what Lexus used to say: “The Relentless Pursuit of Perfection.” That line was rooted in something real — Japanese kaizen manufacturing philosophy, made visible in an ad with a champagne glass balanced on a running engine. You could see the product point. You could feel the benefit it promised.
“Experience Amazing” has nothing underneath it. No mechanism. No proof. No claim that only Lexus could make. Swap it onto any Mercedes, BMW, or Cadillac campaign and change nothing. Not a word.
Asana, the project management platform, is running the same structure on LinkedIn right now. Headline: “Unlock smarter campaign management.” Smarter than what? Measured how? You could replace Asana with Monday.com, ClickUp, Wrike, or Notion and the ad requires zero edits. There is no ingredient, no number, no mechanism. Just a benefit floating in space with nothing holding it up.
This is not a new problem. Bedell named it in 1952. He called them empty claims — benefits announced without the product points that make them believable. And he was precise about what happens when a reader encounters one: they don’t object. They don’t disagree. They simply feel mildly interested and turn the page.
The reader, he wrote, “isn’t sure. She has doubts.”
The Machine That Industrialised the Doubt
Here is what Jasper.ai — the market-leading AI copy tool — asks you to provide before generating your ad:
A product name. A product description.
That’s it. Optional extras: intended audience, tone of voice, background information. What Jasper never asks for: a specific product mechanism, a proof point, a unique differentiator with evidence, or anything that could be described as “why should the customer believe this benefit?”
Writesonic’s Facebook Ads generator asks for three fields: topic, language, tone. Writesonic’s own blog recommends users “include data or leverage an expert voice.” There is no input field in which to do that.
Anyword goes furthest — it lets you select a copywriting framework. AIDA. PAS. BAB. Each generated output gets a Predictive Performance Score out of 100. Sophisticated-looking. But AIDA without a proof point is just a structure for emptiness. The framework selection is a cosmetic layer over the same fundamental gap. No field exists for “what specific mechanism makes this product work?”
All three follow an identical pipeline: product name → description → tone → audience → format → publish.
The step that was always the hardest — the step that required actually knowing what was specific, defensible, and unique about the product — has been quietly removed. Not replaced. Removed.
Prompt templates do the same thing. The widely shared brief format reads: “Act as a Meta Ads copywriter. Write 5 high-converting headlines for , targeting [audience]. Focus on [benefit] and include a call to action.” Not one template includes a step that asks: what specific product point supports this claim?
They are assembly instructions for the Lexus ad. For the Asana ad. For the plum pudding.
"But You've Been Told to Lead With Benefits Your Whole Career"
Every copywriting course. Every conversion framework. Every AI prompt guide. Benefits over features, always. And they’re not wrong.
The problem is the advice stops halfway.
Benefits are the direction good copy travels. Product points are the fuel that gets it there. You need both — woven together, in the right order — not benefits alone with the substance stripped out in the name of brevity or speed.
Bedell tested the difference directly. Selling reprint books by mail, an ad that leaned on benefit language without conviction-building copy pulled 1,000 orders. Rebuilt with the specifics that created genuine belief — same product, same price, same audience — it pulled 2,700. Nearly three times. He opened his chapter on conviction with a line that should be on every content brief written in 2026:
“The need to create conviction is one of the most important factors — and probably the most neglected — in advertising that must produce immediate sales.”
The most neglected. In 1952.
We now have data on what “the most neglected” looks like at industrial scale. Human-written Google Ads are outperforming AI-written ads at 4.98% click-through versus 3.65%. Human-written blogs earn 5.44 times more traffic. Consumer enthusiasm for AI-generated content collapsed from 60% to 26% between 2023 and 2025 — a 34-point drop in two years. Siege Media’s 2025 research found AI-using marketers reporting a 21.5% strategy success rate against 36.2% for non-AI users.
More content. Worse results. The tools are producing more efficiently than ever. What they’re producing more efficiently is doubt.
What the Fix Has Always Been
Bedell rewrote the plum pudding ad. Same product. Same price. Same two-pound portion. Different copy:
“Your family and guests will applaud this wonderful taste treat, with crisp almonds and hazelnuts, plump currants and winy sultanas, fragrant spices and sun-ripened oranges and lemons — all marvelously blended with just enough flour to hold them together. Then the rich whole piquantly flavored with French brandy.”
You can taste it. You can picture it on the table. The reader no longer has to take the advertiser’s word for it — the product points carry the conviction. Not because the writing is more persuasive. Because it gives the reader something to believe.
That’s the gap. Not between bad copy and good copy. Between claimed benefits and earned conviction.
The fix isn’t a better prompt. It isn’t a smarter tool. It isn’t a higher Predictive Performance Score. It’s the work that happens before the brief — before the AI is opened, before a word is written — when someone who actually knows the product sits down and answers a question the tools never ask:
What are the product points?
What’s in this product that only this product has? What’s the mechanism? What’s the specific feature that delivers the promised benefit — and why should the customer believe it without taking your word for it?
If you can’t answer that specifically — not generally, not emotionally, specifically — you don’t have a brief. You have a feeling. And a feeling fed into a content pipeline produces the Lexus ad. The Asana ad.
The plum pudding.
One of those ads sells pudding. Bedell showed us which one in 1952. The data is showing us again right now.
The French brandy was always in the product. Somebody just forgot to put it in the brief.
Find out if your briefs are built to convert — or just built to produce
Most audits we run surface the same problem: campaigns full of benefit language, copy that sounds like marketing, and nobody who’s asked what specific product point supports any of it.
Takes 30 minutes.