Hey, Mark Masters here.

On a Friday afternoon at the end of October I gave Claude a fact sheet for a water shut-off valve that doesn't exist, three competitors included, and asked for a 400-word ad.

The eight ads that came back were good. They had the nine gallons a minute, the holiday leak season, the lab test, the customer who came home to a dry floor, and the free sensor if you order by December 31.

Every one of them also mentioned the one thing none of the competitors could say.

None of them led with it.

Gary Bencivenga had a name for that failure, and a four-part fix he says people have been teaching wrong ever since he first presented it.

I was one of them. Here's what happened when I ran it his way.

Bencivenga's Four Objections, Run Against Claude

You know the Persuasion Equation. Problem, promise, proof, proposition. You probably met it on a slide, the way I did.

For years I used it as an outline. Open on the problem, make the promise, stack the proof, close on the offer.

Then I read what Bencivenga himself wrote about it in his Marketing Maxims, and found out the outline is the part he objects to. The equation tells you what the copy has to answer, in any order you like. It never told you how to arrange the page.

So I built a test around the part that matters, which is the four questions, and ran it on four Claude models. Three of the four came back strong every time.

The fourth is the one your competitors are counting on you to miss.

He Learned Under Caples and Ogilvy, Then Spent Forty Years Beating Controls

Bencivenga's first copy chief was at Prentice-Hall, and the man had no patience for clever. Bencivenga tells the story himself. One pun in a headline and he got the speech every young writer needs once, that the prospect and the product are the stars of the ad and the writer isn't.

He broke into direct marketing working with John Caples at BBDO, and later with David Ogilvy at Ogilvy & Mather, whom he calls his mentor.

By his own account it took him more than ten years to start winning consistently. Once he did, the phone didn't stop. He became the copywriter the big newsletter publishers hired to beat their controls, in split-run tests against other top writers, which is about as close to a gladiator fight as this business gets.

Ken McCarthy calls him the control beater. Brian Kurtz, who spent 34 years at Boardroom, puts him on his Mount Rushmore of copywriters.

Research was one of the three big lessons he says it took him those years to learn. He decided writer's block was really a shortage of material, and he credits Caples with telling him to gather about seven times more interesting information than he could ever use.

In the mid-2000s he retired with a seminar. One hundred copywriters and marketers paid $5,000 each to sit in the penthouse of the St. Regis in New York while he gave away what he knew. He made everyone pledge in writing to keep it to themselves.

That's where he first presented the Persuasion Equation.

Not everyone kept the pledge. The equation spread, his name stayed on it, and the way people taught it bothered him enough that he went public with the correct version.

These days he writes his Maxims, and I'd read every one.

One Antidote for Each of the Four Reasons a Reader Stops

Bencivenga starts from the reader's side. Every ad meets four objections, and each one is a reason to stop reading.

The reader thinks

Bencivenga's name for it

The antidote

"This isn't about anything I care about."

No interest

An urgent problem

"This sounds like every other one."

No difference

A unique promise, one competitors can't match

"I don't believe you."

No belief

Unquestionable proof

"I'll think about it."

No decision

A user-friendly proposition that makes acting now easy

He's specific about three things the slides leave out.

  1. It covers what you say, not how you say it. Wording, design and layout are separate decisions.

  2. The order is up to you. If your offer is the strongest thing you have, test it as the headline. If you have a demonstration that proves the case on sight, lead with it.

  3. You can grade any ad with it. Each part is worth 25 points, and a part that isn't as strong as it could be scores nothing.

Years earlier he'd boiled three of the four down to a single line, in a bullet about reason-why copy. Every ad owes the reader three reasons: "why you, why true, why now?"

Peggy covered the "why true" part with Hopkins in August. This issue is about "why you," because that's where Claude fell down.

Claude Knew the Difference and Put It Halfway Down the Ad

I invented the product: Shutwell Leak Stop, a $279 motorised clamp for your main water valve, with three leak sensors.

The fact sheet gave Claude everything. A split supply line releases 9 gallons a minute.

Most of the leaks customers reported happened between December and February.

An independent lab test clocked the valve shut in 6 seconds, and a customer named Ruth came home from Thanksgiving to a wet cabinet and a dry floor.

Two insurers offer a discount, and there's a free fourth sensor if you order by December 31.

It also listed three competitors.

  • FlowSentry shuts the water off too, but it needs a plumber and Wi-Fi.

  • DripAlert sounds an alarm and can't shut anything off.

  • AquaHalt is a clamp like Shutwell's, but its hub plugs into the wall, so it dies in a power cut.

So Shutwell had exactly one promise nobody else on the sheet could make. It keeps shutting the water off when the power and the internet are out.

Four Claude models (Haiku, Sonnet, Opus and Fable) wrote the ad three ways, twice each, in clean sessions.

One way was the plain request. One added "Use the Bencivenga Persuasion Equation." The third made Claude work through the four objections as separate jobs before it wrote anything.

A separate Claude session graded every ad against the fact sheet, and I checked every headline myself.

Plain request

"Use the Bencivenga Persuasion Equation"

Four jobs first

Led with a promise no competitor could make

0 of 8

4 of 8

7 of 8

Mentioned the outage advantage somewhere

8 of 8

8 of 8

8 of 8

Used the lab test and Ruth's story

8 of 8

8 of 8

8 of 8

Used the December 31 offer

8 of 8

8 of 8

7 of 8

Read the plain headlines with FlowSentry's logo in your head.

  • "Your Pipe Bursts While You're Away. Shutwell Shuts Off the Water in Seconds."

  • "Go away for Thanksgiving. Come home to a dry floor."

  • "Nine Gallons a Minute. Six Seconds to Stop It."

FlowSentry could run every one of them. So could AquaHalt.

Here's the part that got me. All eight plain ads mentioned the outage advantage, and seven compared Shutwell with other devices. It knew. It just put the difference where a feature goes, between word 173 and word 283 of an ad about 400 words long.

Urgency, proof and the offer came back strong every time. The difference, the one element that's specific to this product, was the one Claude filed under features.

My read is that "stops a burst pipe" is the promise this category always makes, and Claude writes the category's promise first. You've seen junior writers do the same thing. I did it for years.

Naming the Equation Got Claude Halfway There

Adding "Use the Bencivenga Persuasion Equation" moved four of the eight ads onto the unique promise, two of them in the headline.

All eight came back in the same order: problem, then promise, then proof, then the offer. Fable labelled its sections "The proof." and "The offer."

To be fair to the equation, six of the eight plain ads were in that order too, so it's mostly Claude's default. But naming the formula didn't change it. Claude reached for the version on the slide, the outline Bencivenga says was never the point, and it got half the drafts to the difference.

Eight Drafts Added a Storm Nobody Put in the Brief

There's a second problem with burying the difference, and it showed up when Claude did try to sell it.

The fact sheet says why leaks are urgent: pipes freeze and people travel between December and February. It says nothing about why an outage might be urgent.

Eight of the sixteen plain and named ads bridged that gap on their own.

❝

Storms knock out power and Wi-Fi, and those are often the same nights pipes freeze.

Claude Opus, plain request

Opus wrote that from the plain request. Fable, Sonnet and Opus wrote versions of it seven more times.

It's plausible, and it might even be true. It isn't in the brief, though, and it's the line that would carry the whole unique promise, which means it's the line your client gets asked about.

The four-jobs prompt added it zero times, and five of its eight runs wrote "no urgency in the facts" next to the outage problem. Peggy caught the same move last month with "make it more specific", and I caught it in the close test two weeks ago. Where the facts run out, Claude writes some.

Run the Four Objections as Four Jobs Before Claude Writes a Word

Here's the prompt as tested, with the paste field added at the bottom.

You're writing a short ad from the fact sheet below. Before you write any copy, do four jobs, one for each reason a reader stops reading: no interest, no difference, no belief and no decision.

1. Urgent problem (no interest). List the problems this product solves that the fact sheet supports. For each one, give the fact that makes it urgent for this reader now, or write "no urgency in the facts". Pick the most urgent.

2. Unique promise (no difference). List every promise you could make. For each one, check the competitors in the fact sheet and name any competitor that could truthfully make the same promise. Keep only the promises no competitor can match. If none survive, write "NO UNIQUE PROMISE" and say what fact you would need.

3. Unquestionable proof (no belief). List every proof item in the fact sheet and rank them from hardest to doubt to easiest, with one line on why. Attach the strongest item to the unique promise. Never add proof the fact sheet doesn't contain.

4. User-friendly proposition (no decision). State the price, the risk reversal and the reason to act now, all from the fact sheet. If there is no real deadline, don't invent one.

Then write the ad: about 400 words, with a headline. Cover all four, in whatever order puts the strongest element first. Use only facts from the fact sheet.

Return your four lists, then a line that says AD, then the ad.

FACT SHEET:
[Paste the product facts, the offer, every proof item you have, and a short entry for each competitor: what it costs, what it does and what it can't do.]

Four things in it do the work.

  1. The promise gets checked against named competitors. "Make it unique" means nothing to a model. "Name any competitor that could truthfully say this" is a test it can run, and it's the step that moved the difference into seven headlines.

  2. Every job has an escape hatch. "No urgency in the facts" and "NO UNIQUE PROMISE" give Claude somewhere to put a gap other than the copy. That's what kept the storm out.

  3. The order is left open on purpose. "Whatever order puts the strongest element first" is Bencivenga's rule, not mine. Three of the four-jobs ads opened on the promise instead of the problem.

  4. The competitor field is the whole game. If you paste in a fact sheet with no competitors, step 2 has nothing to check against, and you're back to the category promise.

It isn't perfect. Haiku ran all four jobs, named the outage promise as the unique one, and then headlined the ad "Leaks Don't Wait for You to Come Home." Read the headline against your own step 2 before you read anything else.

The Prompt Found the Hole in My Own Brief

Six of the eight four-jobs runs told me something about my fact sheet that I hadn't noticed when I wrote it.

The lab test proves Shutwell closes in 6 seconds. Ruth's story proves it works while you're away. Nothing on the sheet proves it works during an outage, and that's the one promise the competitors can't make.

Opus put it plainly: "A test like that from Hartwell would make the lead claim much harder to doubt."

Bencivenga would have sent the brief back for exactly that. Your most unique promise needs your hardest proof, and if a demonstration can prove it on sight, he says lead with the demonstration.

Claude can't run that test for you. It can tell you which test to ask the client for, and that list is worth more than the draft.

Where This Test Is Thin, and What It Can't Tell You

  • One invented product, and I built the difference into it. Real briefs often bury the difference deeper than this, or don't have one, which is a positioning problem the prompt will only report.

  • Two runs per model per arm. That's enough to see a split this wide and too few to put a rate on it.

  • Every grader was Claude. The strict 25-point rule marked down small overstatements like "No subscription, ever," so I've reported the components rather than the totals, and I checked every unique-promise call by hand.

  • Nobody bought anything. Bencivenga built his equation on split-run tests. Mine counts what's on the page.

The Master’s Memo

Claude will cover all four of Bencivenga's objections and still lead with the promise every competitor makes. Make it check the promise against named competitors before it writes, and lead with whatever survives.

If nothing survives, the brief needs work before the copy does.

More clicks, cash, and clients,
— Mark Masters