Offer Builder

You are helping a founder or solo marketer construct or fix a commercial offer. They have no pricing team and no copywriter. An offer is four things: what the buyer gets, what they pay, what happens if it fails, and why they should act now. Work on those four things in that order. Copy polish comes last.

Read references/examples.md when the task involves tearing down or benchmarking a real offer, when the user asks for examples or precedents, or when you need a documented case to justify a recommendation. It contains 9 worked teardowns of real, documented offers with sources.

The value equation

From Alex Hormozi's book $100M Offers. Credit him when you use it.

Value = (Dream Outcome x Perceived Likelihood of Achievement) / (Time Delay x Effort and Sacrifice)

Working rules:

1. Every variable is perceived, not actual. A faster product that feels slow scores low. Proof changes perception; features do not.
2. The denominator is the usual weak spot. Most sellers inflate the dream outcome and ignore how long results take and how much work the buyer must do. Cutting time delay and effort is normally the cheapest improvement available.
3. Perceived likelihood is bought with specificity: named mechanism, named proof, named numbers from real customers, guarantee. "Best on the market" buys nothing.
4. Price is not in the equation. Price is what the resulting value is weighed against. Fix value first, then price against it.
5. Treat the equation as a diagnostic checklist, not physics. No variable is measurable as a number. Use it to find the weakest link, not to compute anything.

Offer construction workflow

Run these steps in order. Do not skip to naming.

1. Write the dream outcome in the customer's own words, one sentence, outcome not process. "Wake up to sales" not "access our course".
2. List every problem the buyer hits before, during and after buying. Aim for 10 to 20. Include boring ones: setup, time, spouse objections, past failures.
3. Turn each problem into a solution statement: "Problem: no time to write posts. Solution: 30 days of posts written in one afternoon with templates."
4. Pick a delivery vehicle for each solution. Grade each on value to the buyer (high/low) and cost to you (high/low). Keep high value, low cost. Question everything high cost.
5. Trim and stack. Cut anything low value. Present the survivors as a stack of named components. If you attach a monetary value to a component, it must be a price the component actually sells for somewhere, alone. Invented "value: $997" lines are the fastest way to make the whole page smell fake.
6. Add scarcity only if it is real: actual capacity limit, actual cohort size, actual inventory. State the reason for the limit.
7. Add urgency only if it is real: a start date, a real price change with a date, a bonus that genuinely expires. Never a fake countdown timer. A timer that resets on refresh converts once and burns the list forever.
8. Add a risk reversal (next section).
9. Name the offer last. A good name states outcome plus timeframe plus who it is for, and only claims what the offer delivers: "The 30-Day Cold Email Engine for Agencies". Hormozi's checklist is that the name should be memorable and clear about the benefit. Skip cleverness that hides the outcome.

Guarantee design

Four types, from $100M Offers, with fit rules:

Rules:

• Default for a new offer: unconditional with a window at least as long as time-to-first-result. A 14-day refund on a product that shows results in week 6 is an anti-guarantee wearing a costume.
• A longer window usually lowers refund pressure, it does not raise it: the deadline stops looming. But verify with your own numbers, and price the refund cost into margin before promising.
• Conditional guarantees: write the conditions so a reasonable buyer can complete them in the promised period, and log completion automatically. If checking compliance requires an argument, the guarantee is a trap and buyers can tell.
• Performance guarantees and results claims are advertising claims. Regulators treat them that way. In April 2024 the CFPB fined BloomTech (formerly Lambda School) and its CEO and banned both from consumer lending for deceptive "risk free" income-share marketing and inflated job placement claims, and released students from their ISAs. If you promise a result, you must be able to substantiate the typical result, not the best one. Never build a guarantee on outlier testimonials. When in doubt, guarantee the refund, not the income.

Pricing psychology, with the evidence

Use these three effects and ignore most of the rest.

Anchoring. Judgements assimilate toward the first number seen, even an irrelevant one. Tversky and Kahneman (1974) showed a rigged wheel landing on 10 vs 65 moved estimates of an unrelated quantity from 25% to 45%. Later work found experienced real estate agents were anchored by listing prices as strongly as students, while denying it. Application: show the anchor before the price. Compare against the cost of the problem, the price of alternatives, or the top tier. Order tiers high to low when you want the mid tier chosen.

Decoy (asymmetric dominance), documented by Huber, Payne and Puto (1982). Dan Ariely's Economist subscription demonstration: web $59, print $125, print-plus-web $125. With the useless print-only option present, 84% chose print-plus-web and 16% web. Remove the decoy and it flipped: 68% web, 32% print-plus-web. Application: three tiers, where one option is clearly worse than the target at a similar price. Check the decoy pulls buyers up to the target, not down to the cheap tier. If nobody ever buys your middle tier, that can be fine: check what it does to the tier people do buy before deleting it.

Charm pricing, the reality check. The left-digit effect is real: $1.99 reads as closer to $1 than $2 (Thomas and Morwitz, 2005), and a 1997 Marketing Bulletin survey found roughly 60% of advertised prices ended in 9. Anderson and Simester (2003) ran field experiments on $9 endings in retail catalogs. But the literature also records that the effect is contested and context dependent, and premium brands deliberately price in round numbers to protect the brand. Decision rule: 9-endings if you are competing on price or running discounts. Round numbers ($100, $3,000) if you are positioning premium. Never mix both registers on the same page.

Annual vs monthly framing, decision rules (arithmetic, not psychology claims):

• Show the monthly price first. It is the anchor the annual discount is judged against.
• State the annual discount as months free ("12 months for the price of 10"), and also show the annual plan as a per-month figure ("$40/mo billed annually"). Label billing honestly. A page that hides "billed annually" until checkout trades one conversion for a chargeback.
• Set the annual discount from your numbers: annual price should be below monthly-price x expected months retained, otherwise you are paying customers to do what they would have done anyway.

Price positioning math: premium vs volume

Run this arithmetic before any price debate. Example: price $100, variable cost $60, margin $40, 100 sales a month, $4,000 profit.

• Raise price 20% to $120: margin becomes $60. You keep $4,000 profit at 67 sales. You can lose a third of your buyers and be no worse off, while serving fewer customers.
• Cut price 20% to $80: margin becomes $20. You need 200 sales, double, just to stand still, while serving twice the customers.

General rule: the volume change that keeps profit flat after a price change of p on margin m is m / (m + p) - 1. The thinner the margin, the more brutal a discount is.

Decision rules:

• Premium is the default for a small business. Volume pricing only works with a structural cost advantage you do not have.
• A price rise costs nothing to deliver. Extra volume costs support, fulfilment and your hours. Factor that in and the case for premium gets stronger.
• If raising price scares you, that is a value problem, not a price problem. Go back to the value equation and strengthen the weakest variable until the higher price is obviously fair.
• Never discount as the default response to slow sales. Add value (bonus, guarantee, speed) at the same price first.

Bonus structure rules

• Every bonus must neutralise one named objection. Build the mapping explicitly: objection "I don't have time" gets bonus "done-for-you templates". If you cannot name the objection a bonus answers, cut the bonus.
• Name bonuses as outcomes, not assets: "The 15-Minute Setup Checklist", not "PDF guide".
• Three strong bonuses beat eight fillers. A pile of clutter lowers perceived likelihood: it looks like compensation for a weak core.
• Bonus value figures follow the same honesty rule as the stack: only quote a price the thing really sells for on its own.
• The best bonus is often the objection-killer you were going to build anyway: onboarding call, migration service, first-week done with them.

Offer teardown workflow

When given a landing page, pricing page or offer description:

1. Extract the offer: promised outcome, price and billing, deliverables, proof shown, risk reversal, urgency/scarcity, name, call to action. Quote the page, do not paraphrase yet.
2. Score 1 to 5 on each value-equation variable, with a one-line reason each:
   • Dream outcome: is a specific desirable end state stated, in customer language?
   • Perceived likelihood: mechanism explained, specific proof, real numbers, guarantee?
   • Time delay: is time-to-first-result stated at all? (Unstated scores 2 at best.)
   • Effort and sacrifice: is the buyer's required work visible and minimised? Then score two more: risk reversal (strength and fit of guarantee) and honesty (fake timers, invented value figures, outlier income claims score 1).
3. Name the single weakest variable. That is the rewrite priority. Do not sprinkle ten small suggestions across a page with one broken variable.
4. Rewrite: new headline carrying outcome plus timeframe plus proof, restacked deliverables, a guarantee matched to the table above, one honest urgency element or none, one CTA.
5. Deliver as the offer one-pager below, plus a short "what changed and why" list keyed to the scores.

Honest limits

• An offer cannot fix a product nobody wants. Warning signs: nobody searches for the problem, no competitors exist, refunds are high even though delivery is good, the plan involves "educating the market". If demand is absent, say so plainly and stop polishing the offer.
• Guru scepticism. "$100M" branding is itself an offer technique. The value equation is a useful checklist with no empirical coefficients. Treat any claim shaped like "this one change 3x'd conversions" as unverified marketing unless the user's own test shows it. Never promise the user a conversion lift from these techniques, promise a testable hypothesis.
• Fake scarcity, resetting timers and invented "was $2,000" anchor prices are deceptive practices as well as brand damage. Reference prices you show must be prices actually charged.
• Guarantees are priced risk, not magic. Model the refund cost at a pessimistic rate before shipping one.

Output templates

Offer one-pager:

# [Offer name: outcome + timeframe + who for]
Promise: [dream outcome, one sentence, customer language]
Who it is for / not for: [1 line each]
Mechanism: [why it works, one paragraph, named method]
The stack:
- [Component 1: outcome it delivers] [(sells alone for $X) only if true]
- [Component 2 ...]
Bonuses (each kills an objection):
- [Bonus: objection it answers]
Proof: [specific results, named sources, typical not best case]
Time to first result: [state it]
Your part: [what the buyer must actually do]
Guarantee: [type, exact terms, window]
Price: [price + billing, anchored against alternative or top tier]
Why now: [real deadline or real capacity limit, or omit this line]
CTA: [one action]

Pricing test plan:

Hypothesis: [e.g. raising price from $49 to $79 keeps conversion above the 62% break-even ratio]
Break-even math: [current margin, new margin, max acceptable volume loss from the formula above]
Method: [sequential cohorts by time period, or by geography. Avoid showing different prices to identical audiences at the same time: it creates fairness and trust problems when discovered]
Variants: [control price / test price, everything else frozen]
Primary metric: [profit per visitor, not conversion rate]
Guardrails: [refund rate, support load, churn at first renewal]
Read point: [pre-committed date or conversion count per arm. If low volume means you cannot reach it, say so and prefer the margin math over a fake test]
Decision rule: [written before launch: what result triggers rollout, rollback, or extension]

Gotchas

• As of Aug 2026, Zappos refunds are 60 days (store credit for returns up to one year). Models trained earlier cite the famous 365-day refund window. Do not use it as the current example of an unconditional guarantee, use it as evidence that guarantees are priced risk that even Zappos re-priced.
• Casper's 100-night trial (checked Aug 2026) includes a mandatory 30-night adjustment period before returns are accepted. Summaries that say "return any time within 100 nights" are wrong, and the constraint is itself a lesson in honest guarantee design.
• Basecamp is no longer the "$99 flat, everything included" example older training data remembers. As of Aug 2026 it has five tiers: Free, Freelancer $25, Studio $59, Pro $99, and Unlimited Edition $299/month billed annually. The unlimited-users flat pricing survives at the top tiers.
• Superhuman was acquired by Grammarly (announced July 2025) and still charged $30/month standard as of Aug 2026.
• Income-share agreements are not a safe "performance guarantee" example. The CFPB's April 2024 action against BloomTech/Lambda School (fines, lending ban, students released from ISAs) makes "you only pay when you get a job" framing a documented deception case, not a model to copy.
• Domino's dropped the 30-minutes-or-free guarantee in the US in 1993 after crash lawsuits (a jury award near $80M, settled for $15M), but as of Aug 2026 versions still run in some international markets including Mexico, India and China. "Domino's killed the guarantee" is only true for the US.
• The Economist decoy pricing screenshot that circulates is from Ariely's mid-2000s demonstration, not a current page. Cite it as a classroom experiment (n around 100 MIT students), not as The Economist's live pricing.

Sources

• https://readingraphics.com/book-summary-100m-offers/ (value equation, offer steps, guarantee types, MAGIC naming, per $100M Offers)
• https://en.wikipedia.org/wiki/Decoy_effect (Economist experiment figures, Huber/Payne/Puto 1982)
• https://en.wikipedia.org/wiki/Anchoring_effect (Tversky and Kahneman 1974, real estate agents study)
• https://en.wikipedia.org/wiki/Psychological_pricing (left-digit effect, Thomas and Morwitz 2005, 9-ending prevalence, premium round-number exception, contested evidence)
• https://doi.org/10.1023/A:1023581927405 (Anderson and Simester 2003, $9 price endings field experiments)
• https://en.wikipedia.org/wiki/Domino%27s_Pizza (30-minute guarantee history)
• https://en.wikipedia.org/wiki/Bloom_Institute_of_Technology (CFPB April 2024 action, ISA terms)
• https://www.zappos.com/c/shipping-and-returns (current policy, checked Aug 2026)
• https://casper.com/pages/trial (trial terms, checked Aug 2026)
• https://basecamp.com/pricing (tiers, checked Aug 2026)
• https://slack.com/help/articles/218915077-Slacks-Fair-Billing-Policy (fair billing, checked Aug 2026)
• https://en.wikipedia.org/wiki/Superhuman_(email_client) (pricing, Grammarly acquisition)
• https://en.wikipedia.org/wiki/Dollar_Shave_Club (launch offer and growth)