Pricing Page Teardown

Pricing Page Teardown

Most pricing-page reviews are opinions about layout. Most price tests are launched by teams who find out afterwards that they never had the traffic to learn anything. This skill replaces both with arithmetic: 21 checks scored on what a buyer can actually tell from the page, and a test plan that says, before anything ships, whether a live test can answer the question in the weeks you have. For Fieldnote, the invented scheduling app in the worked example, a $49 → $59 test at 2% conversion and 3,000 visitors a week needs 126,757 visitors per arm, which is 85 weeks. That result is the point: it is not unusual.

When to use

The user is reviewing their pricing page, comparing it with competitors, considering a price change, or has a price test running and wants to read it.

Do not use it to design the tiers or the value metric; pricing-and-packaging does that. Do not use it to estimate willingness to pay; the Price Sensitivity Analyzer asks the four price questions and gives the acceptable range.

Before you start

1. Look at the pages as a buyer, today. Answer each check on what is visible now, not what is planned. "Not sure" scores zero; go and look.
2. For a test, put the two prices on the same basis (first month, first year, or first invoice) and pull conversion from at least a month of your own analytics.
3. Write an honest guess at conversion at the test price. The break-even line tells you how far it can fall before the new price loses revenue.
4. Count the weeks you could keep a test clean. Eight is AAJ's rule of thumb.

Method

Twenty-one checks in five areas - clarity, choice, proof and risk, friction, honesty and terms - each weighted 1 to 3 by what it costs a buyer (AAJ's judgement, not a benchmark). Yes scores 2 points × weight, partly 1, no and not sure 0; unanswered checks are left out. Each page gets a share of the weighted points available on the checks answered, and your fixes are ranked by points lost and then by how many competitors already pass them.

The honesty checks come first whatever they score. In the US, ROSCA requires clear disclosure of all material terms before billing information is taken, express informed consent, and a simple way to stop recurring charges. California's automatic renewal law requires online cancellation for online sign-ups and 7-30 days' notice of a change in the amount charged (contracts from 1 July 2025). The FTC's Guides Against Deceptive Pricing treat a fictitious former price as a false bargain. The engine flags any of checks 18-20 that fail.

Plan the test on revenue per visitor, not conversion. Break-even conversion at the test price is today's conversion × current price ÷ test price. The visitors needed per arm come from a two-sided test on revenue per visitor at your alpha and power (0.05 and 0.8 by default), and are converted to weeks at your traffic. The engine also reports the smallest change your traffic could detect in the weeks you have. Then:

• testable: the weeks needed fit in the weeks you have. Run to the full size; never stop early because it looks good;
• not testable: use an alternative - ask before you charge, change it for new customers and compare whole-week periods, quote it in sales-led deals, or test a bigger change (the sample shrinks with the square of the effect);
• guess says no: your own guess already has the test price losing revenue;
• nothing to detect: both guesses give the same revenue per visitor.

Read against the plan. Conversion, revenue per visitor and a two-sided z-test p-value per arm. Below the planned size the read-out is "too early", whatever the p-value says. Before-after and sales-quote designs are marked not randomised and read as direction. A 90-day retention check per arm is asked for before any price becomes permanent.

Workflow

Step 1 - Score. Your page and up to three competitors on the 21 checks.

Step 2 - Fix. The honesty checks first, then the top five by points lost.

Step 3 - Plan. Run the test plan. If it is not testable, pick the alternative and write down its weakness.

Step 4 - Decide the rules before launch. Keep the test price if revenue per visitor is higher with p below alpha at the full size; keep the current price otherwise; stop early only on a guardrail (refunds, cancellations, discount requests, support tickets about price).

Step 5 - Read. Give the engine both arms and the planned size. Then the retention check at 90 days.

Run the tool

node resources/pricing-teardown.js --demo

run({ teardown, testPlan, testRead }) returns the scored pages with area shares and ranked fixes, the plan with break-even, visitors per arm, weeks, minimum detectable change and verdict, and the read-out with p-value, progress against plan, verdict and retention note.

Present the result

1. The scores, yours against each competitor, and the area where you lose most.
2. The top five fixes, with the honesty checks called out first.
3. The test plan in one sentence: break-even, visitors per arm, weeks, and the verdict.
4. The read-out against the planned size, and whether retention has been checked.

Guardrails & common mistakes

• Never test on existing customers. Their prices change with notice, never as an experiment.
• Never stop a test because it looks good. Checking repeatedly and stopping at the first significant reading inflates false positives.
• Never read conversion alone. A price that attracts trials can still lose revenue.
• Do not use a former price you never charged. It is a false bargain under the FTC guides.
• I am not a lawyer. The legal checks point at the rules; check them, and local rules outside the US, before relying on any of it.

Sources

• 15 U.S.C. § 8403, Restore Online Shoppers' Confidence Act - disclosure, consent and cancellation for negative-option features.
• Cal. Bus. & Prof. Code § 17602 - automatic renewal: notice of a change in charge; online cancellation.
• 16 CFR § 233.1, Former price comparisons - FTC Guides Against Deceptive Pricing.
• NIST/SEMATECH e-Handbook, 7.2.2.2 Sample sizes required - the sample-size formula.
• Evan Miller, How Not To Run An A/B Test - on stopping early.

Related AAJ resources

• The workbook this skill runs, with the competitor grid and a 30-row test log: https://aajconsult.com/resources/pricing-page-teardown
• Asking before you charge: https://aajconsult.com/tools/price-sensitivity-analyzer
• What a price change does to ARPU: https://aajconsult.com/tools/pricing-arpu-modeler
• Designing the tiers in the first place: https://aajconsult.com/playbooks/pricing-packaging-playbook

Related skills

• pricing-and-packaging - the tiers and the value metric the page presents
• ab-test-significance - the same test maths for conversion rather than revenue per visitor
• website-conversion-audit - the rest of the page around the price table
• aaj-marketing-psychology - anchoring and choice architecture, and where persuasion ends