pricing-test

name	pricing-test
description	Designs a pricing experiment — the hypothesis with its evidence, one structural change (band, anchor, or packaging), a run plan that protects trust, and the decision rule written before launch. Use when the price was set by guesswork and it is time to learn what the market actually pays.
argument-hint	[current price + what you suspect + what you sell]

Pricing Test — find the price without burning the market

Most prices are archaeology: set once at launch, defended forever. Yet price is the fastest profit lever in the business — a rise flows almost straight to margin — and it goes untested because a botched test torches trust. The fix is not bravery, it is design: small surface, one change, clean read, decided in advance.

Inputs

• Current price and model, what you suspect is wrong with it, and what is being sold: $ARGUMENTS
• From marketing-brief.md if present: the offer (price and its logic), numbers that matter (conversion at the current price, acquisition cost), objections & beliefs (price objections), and any willingness-to-pay evidence — deals lost on price, and deals that closed suspiciously easily. Closing too easily is the under-pricing signal everyone ignores.

Do this

1. Establish the baseline: current price, close rate or conversion at that price, and revenue per prospect. This is the number the test must beat.
2. Write the hypothesis: "We believe [segment] will accept [price or structure] because [evidence]." No because, no live test — run willingness-to-pay research first. A live test is the expensive way to ask what research answers cheaply.
3. Choose ONE structure to change:
   • Band test — the same offer at a higher price, shown to a new cohort only.
   • Anchor test — add a premium tier whose job is to reframe the core price.
   • Packaging test — the same total value re-split so old comparisons break.
4. Design the trust protection: new prospects only — existing customers are never repriced mid-relationship without being told; every quoted price gets honoured; decide up front what test-cohort buyers get if the test price wins or loses. No fictional strikethrough "was" prices.
5. Define the read: enough purchase decisions per cell to believe the result — rough sanity maths, not false precision. Low-volume businesses read sales calls and close-rate shifts, not dashboards. The deciding metric is revenue per prospect, not close rate.
6. Write the decision rule before launch: "If revenue per prospect is at or above X after N decisions, the new price is the price. Below Y, revert. Between the two, extend once, then decide." Put the review date in the calendar.
7. Log it in marketing-brief.md: the test card goes into Learnings & experiments now; when the test decides, update the offer section's price and its logic — the brief's price is the real one.

Output

The pricing-test one-pager: baseline, hypothesis with its evidence, chosen structure and why, trust rules, read plan with the sample logic, decision rule, start and review dates.

Rules

• Never test on existing customers unannounced. One trust breach costs more than a year of price optimisation earns.
• Revenue per prospect decides, not close rate — a lower close at a higher price often makes more money.
• One structural change per test. Price and packaging moved together produce an unreadable result.
• Evidence over invention: if willingness-to-pay evidence is thin, say so and gather it first rather than dressing a guess as a hypothesis.
• Anchor tiers must be real products someone can buy. A decoy you would refuse to sell is a lie with a price tag.