Hi, it's Peggy Burnett.
I gave eight separate Claude sessions the facts behind a post-purchase upsell and asked each one to write the page.
Five of them got a plain brief, and all five made the same two choices:
They admitted the company's own motive for the offer, four of them with the same opening sentence.
None of them closed by pointing back to the purchase the buyer had just made.
That missing close is the line I would defend hardest on the page this issue takes apart, a $2,000 onboarding upsell Kleo showed its buyers in August. It is the final headline on the page, and it applies less pressure than anything else on it.
I expected Claude to fail on pressure, stacking countdowns and scarcity the way a lot of upsell pages do. The count put the failure somewhere else. Claude was, if anything, too polite.
A $2,000 Upsell Page, Taken Apart Line by Line
Kleo makes software for writing LinkedIn content. When Kleo 3 launched in August, buyers who checked out were shown this page before they reached their account.
It offered White-Glove Onboarding for $2,000, sold on top of a year of the software and a four-week group cohort called Momentum.
I am working from a full-page capture taken in August, and the offer closed on 12 August, so nothing here is a live pitch.
The page also carries a five-minute video. I have not analysed it, because I can only count what I can read.
Before reading it for quality, I counted what is on it.
Element | Count |
|---|---|
Buy buttons | 4 |
Statements of a deadline or a loss of access | 6 |
Uses of "fast," "fastest," "fast-track" or "fast-lane" | 10 |
Uses of "ROI" | 2 |
Figures given for the ROI or the speed | 0 |
The Headline Spends Fifteen Words and the Subhead Does the Work
The headline reads: "Fast track your success with Kleo today: get done-with-you implementation with our premium White-Glove Onboarding." A yellow badge beside it adds "FASTEST ROI!"
That is fifteen words if you count the hyphenated compounds as one word each, and eighteen if you split them. It has two claims with nothing behind them, and one of those is the brand describing its own product as "premium."
The subhead underneath is the better piece of writing: "A done-with-you onboarding built for teams and agencies who don't need more education, they need Kleo set up right, and running, from day one."
That sentence names who the offer is for and what they want, in the buyer's terms. If I were rewriting the top of this page, I would promote it to the headline and cut the badge.
The Confession Buys Credibility, and the Next Paragraph Spends It
The block that makes this page worth a teardown sits below the price card, under the heading "I have something to confess..."
Part of the reason we're doing this premium White-Glove Onboarding is because we didn't want to host more education calls. Why? Because at your level, you don't need more education, and spending hours on calls doesn't make either of us more money.
That is a company telling a buyer it prefers this option partly because it pays better. It is also the reason-why principle I wrote about in the Hopkins issue in August, applied to the offer itself.
A buyer who has just spent money is primed to wonder why am I being sold to again? This block answers the question before they can ask it.
The paragraph that follows gives much of it back.
While more education calls aren't as valuable, what is extremely valuable, especially for agencies and teams, is a done-with-you white-glove service, where you can fast track your success with Kleo and get the fastest ROI on the platform.
"Extremely valuable" and "fastest ROI" are the same unsupported claims as the headline badge.
The confession earned the page a moment of belief, and the next sentence spent it on an assertion.
Phases A, B and C Make a Service Easy to Picture
The deliverables are the page at its most concrete. Nearly every item in the three phases carries a number or a named thing you receive.
Before the call: a 3-minute intake form, your top three past LinkedIn posts, and team seats set up by Kleo's staff.
The call: 45 to 60 minutes, live, with Jake and the Kleo team.
After the call: the recording, a custom PDF SOP, and 30 days of priority chat support, plus a bonus call usable within three months.
A sequence you can picture is its own kind of proof, and it does more than the ten uses of "fast" put together.
The VIP Chat section that follows is the weakest writing on the page, for two reasons.
It promises "6 and 7-figure entrepreneurs" without saying how many, and its whole argument is "your network is your net worth."
One benefit card reads "No forever membership. Access comes as part of this upsell, not another ongoing commitment." I read that twice and still cannot tell whether access ends.
A buyer about to spend $2,000 should not have to guess.
The Page Carries Two Deadlines That Disagree
Six statements on the page tell the buyer when the offer ends. They sort into two groups that contradict each other.
Says the buyer has days | Says the offer vanishes on leaving |
|---|---|
A countdown reading two days and thirteen hours | "Careful, don't click away or refresh this page" |
"This offer closes 12th Aug," printed three times | "Not shown again after checkout" |
A buyer cannot know which column is true, and six deadline statements add up to less certainty than one clear date would give.
The ordering has a similar problem. The note from Jake opens with "First off, congratulations and thank you."
It appears below the headline, the price, the full offer card and a buy button, so the first thing said to the buyer is the second offer.
The Closing Panel Argues From a Decision the Buyer Already Made
The page ends on a purple panel.
You already decided Kleo was worth 12 months of your time. This just makes sure those 12 months start on the right foot, instead of the slow one.
Every argument above it asks the buyer to make a new decision about a new purchase. This one asks them to protect a decision they already made.
The question it puts to them is whether the year they just bought should start well, and they answered that a minute ago when they paid.
The strongest argument on the page is the only one that asks nothing new of the buyer.
The decline link is the other piece I would keep: "No thanks, I'll build it myself with the cohort."
It names a real alternative the buyer already owns, in their own voice, with no guilt attached. That matters here, because a buyer shamed one minute after purchase starts reconsidering the first purchase too.
Claude Wrote the Confession Every Time and Never the Close
To see which of these moves a model reaches for on its own, I built a fictional offer with the same shape and gave it to Claude.
Brightledger is bookkeeping software for small agencies. The buyer has just paid $588 for a year, including a four-week group cohort. The upsell is a $1,200 done-with-you setup built around a 60-minute private call, available for 72 hours.
I also gave Claude the business reason, including the part that benefits the company: the group calls fill up with setup questions and are expensive to run.
Kleo's writer had to decide to disclose a motive like that, and Claude was simply handed it. So any confession it wrote shows what it did with a reason it was given, and whether it would have volunteered one is a separate question.
Five fresh sessions each got the brief with nothing else, and I counted the same elements on all five pages.
On the page | Cold drafts (of 5) |
|---|---|
Purchase confirmed before the price appears | 5 |
Company's reason for the offer stated | 5 |
Company's own cost admitted in plain words | 3 |
Reason opened with "We'll be straight with you" or a close variant | 4 |
Countdown timer | 1 |
Deadline stated more than once | 4 |
Decline link names the cohort the buyer already owns | 5 |
A section telling some buyers to skip the offer | 4 |
Closes by pointing back to the purchase just made | 0 |
At least one fact the brief did not contain | 5 |
I expected the pressure layer, and it barely appeared: one countdown in five pages and no refresh warnings. Four other patterns showed up instead.
It Talked Buyers Out of the Offer
Four of the five included a section explaining who should decline. One told readers outright: "If you're in the right-hand column, decline this offer."
That is a defensible choice on a $1,200 add-on. It is also a choice the model made for the client without being asked.
It Sanded Down the Confession and Wrote It From a Template
Two of the five drafts dropped the admission that the group calls cost money and kept only the part about serving the buyer faster.
The wording converged as well:
Four opened the reason with "We'll be straight with you" or something very close to it.
All five put it under a heading that began "Why We're Offering This."
Five sessions converging on one sentence is how a reader spots template output.
It Invented Facts About a Product That Does Not Exist
Every cold draft added at least one fact I never supplied.
Four described how "most agencies" use the bonus call.
Two said the cohort runs every month.
One said a buyer gets five minutes on a group call.
One said the calls are shared with "fifteen other agencies."
None of those is in the brief, and Brightledger has no customers to generalise from.
It Never Closed on the Decision
None of the five pointed back to the purchase the buyer had just made. Three ended on the timing of the offer, one on a list of what happens after you click, and one on a short FAQ.
A Prompt That Asks for the Close Gets It, Plus a New Stock Sentence
The fix I tested makes the model state three things before it writes:
What the buyer just committed to.
What the company gets out of the offer.
What the buyer loses by declining.
Three more fresh sessions got this prompt with the same Brightledger facts.
You are writing a post-purchase one-time-offer page. The reader bought from us less than a minute ago and has not used the product yet. Everything on this page either confirms that decision or makes them doubt it.
Work in this order.
1. Before writing any copy, answer three questions in two or three sentences each.
(a) What did the buyer just decide? State it as a commitment of their time or money, in their terms, rather than as a product name.
(b) What is our real reason for making this offer now? Include the part that benefits us.
(c) What does the buyer lose by declining, given that what they already bought still works? Be honest if the answer is "not much."
2. Write the full page, top to bottom, including headline, subheads, body, button text and decline link. Follow these rules.
- Confirm the purchase in one short line above the headline, before the price appears anywhere. The headline itself is about the offer.
- State the reason from 1(b) in plain words, including our own interest. Start on the fact itself. Do not introduce it with a sincerity phrase such as "We'll be straight with you," "Here's the truth" or "I have something to confess."
- Close on the decision from 1(a). The last argument on the page should be about getting full value from what they already committed to.
- Write the decline link as the real alternative they already own, in the buyer's voice, with no guilt or loss framing.
- State the deadline once, beside the final button, and only if it is real. No countdown timers, no refresh warnings, no repeated deadline lines.
- Every outcome claim must rest on a deliverable or a number in the offer facts. No ROI, speed or results claims the facts do not support.
3. After the page, list every sentence that claims a benefit. Next to each, name the fact in the brief it rests on. Flag any sentence that rests on nothing, then rewrite or cut it.
Offer facts:
[Paste what the buyer just bought and what they paid, every deliverable in the upsell with its numbers, the price, the real deadline, what the buyer keeps if they decline, and your honest reason for making the offer.]All three guided drafts did what the rules asked.
Each closed on the purchase.
Each admitted the company's cost in plain words, with no sincerity phrase in front of it.
None used a countdown, and each stated the deadline once.
Step three did the most work. The audit caught two, three and four unsupported sentences in the three drafts, nine in total, and each session cut or rewrote them before handing the page back.
The audit still leaves some checking to you. One sentence that a sibling session flagged as an inference survived in another.
Two costs came with the prompt, and I have fixed only one of them.
The Confirmation Took Over the Headline
My tested wording said only "confirm the purchase before the price." All three sessions read that as an instruction to make the confirmation the headline.
"Your Brightledger Account Is Live" is a good receipt and a poor headline for a $1,200 offer. The version above adds a line that keeps the headline on the offer, and that line is untested.
All Three Closes Came Back as One Sentence
The prompt put a close on every page, and all three came back nearly identical.
Session | Opening of the closing section |
|---|---|
1 | "You've committed $588 and a year of your agency's books to Brightledger." |
2 | "You've committed $588 and a year of your agency's bookkeeping to Brightledger." |
3 | "You just committed $588 and a year of your agency's books to Brightledger." |
The prompt moved the convergence from the confession to the close.
Each of those sentences is accurate, and none of them has the tension of "start on the right foot, instead of the slow one." Kleo's line works because it sets the buyer's decision against a specific, slightly uncomfortable alternative.
The prompt can make sure a close exists and points in the right direction. In eight drafts, the model never wrote one I would ship unchanged.
Where Eight Drafts on One Fictional Offer Stop Telling You Anything
Three limits apply to everything above.
This is one page, captured once, from one company. I have no conversion data for it, so when I call a line strong, that is a craft judgment.
The Claude side is eight drafts across two conditions, on one offer I invented, from one model. That is enough to see a habit and too few to measure a rate.
The brief handed Claude the company's motive. A model given only the deliverables and the price might never write a confession at all, and I did not test that.
What transfers is the reading order, and you can run it on your own upsell page.
Count the claims that have nothing behind them.
Check whether the deadlines agree with each other.
Find the sentence that argues from the decision the buyer already made.
If your page does not have that sentence, write it yourself. None of my cold drafts wrote it unprompted.
The Burnett Matrix
I went looking for a model that oversells and found one that under-closes.
Claude will admit your motive, respect the buyer's exit and trim its own claims when you ask it to audit them. It will write a closing line only when you tell it where the close should point, and even then the line comes back sounding the same every time.
That line is still the copywriter's job.
More opens, clicks, and conversions,
— Peggy Burnett
