Pricing Experiments Catalog: 12 Rapid A/B Tests to Validate Billing Units in 30 Days

Before running tests, define baseline ARPU and the cohort you’ll measure (new signups, trialers who reach paywall, or existing customers). Use revenue-per-user over a fixed window (first 30 days or first billing cycle) as your primary conversion-to-ARPU mapping — it’s direct and easy to compute. Measure both conversion rate and revenue-per-converted-user so you can calculate ARPU change = conversion_rate * avg_revenue_per_payer.

Pick conservative minimum detectable effects (e.g., 5–10% relative lift in conversion or ARPU) and use a sample-size calculator (your A/B tool or common calculators) to avoid underpowered tests. If traffic is limited, run sequential pilots on segmented cohorts rather than full randomized A/B tests to reduce risk and time to learning. Keep tests isolated to new customers or a clearly segmented subset to avoid cross-cohort billing confusion.

The next 12 templates are grouped into three families: Billing unit swaps (per-seat, per-user, per-output, per-project), Add-on & microfeature tests (one-time, token packs, feature unlocks), and Packaging & framing (annual prepay, contrast pricing, free-trial duration). Each template is compact so you can ship in a sprint with conservative guardrails.

For every template: run duration 14–30 days, primary cohort = new paying-eligible signups (or trial converters), and capture: conversions to paid, revenue per payer, churn in first 30 days (if measurable). Use feature flags or your experimentation system tied to billing (RevenueCat/Lago/your billing provider) to keep revenue telemetry accurate and reversible.

1) Per-seat vs per-active-user. Hypothesis: switching to per-active-user reduces sticker shock and increases conversion among small teams. Sample: randomize new signups into two variants, each with n determined from expected baseline conversion and 7–10% MDE. Metric: conversion rate and revenue/payer over 30 days. Implementation: show variant pricing on paywall and route billing to same plan ID logic but compute invoices based on reported active users. Rollback: restore control and refund any misbilled invoices within 48 hours.

2) Per-output (credits) vs flat plan. Hypothesis: charging per-output (credits) aligns payment to value and increases conversion for low-frequency users while increasing ARPU for heavy users. Implementation needs ledger-backed credit balances and per-charge events; run on new users only. ARPU mapping: compute ARPU change by modeling average credits consumed per cohort times credit price.

3) Microcheckout (one‑click microfeature). Hypothesis: offering a one-time microfeature purchase at the first high-value moment yields measurable paid adoption and predictable ARPU lift. Sample: show the microcheckout to 50% of qualifying users who reach the key moment. Metric: microfeature conversion rate, revenue per microfeature purchaser, and downstream retention. Implementation: simple 1-click checkout integrated with billing; if using token packs, deliver tokens to user ledger.

4) Token packs and bundles. Hypothesis: selling token packs (3/10/30) with a rollover increases ARPU and average spend-per-buyer. Implementation: offer token pack variants and measure purchases and average tokens consumed. Rollback: pause new purchases and honor already delivered tokens or refund unused portion per your policy.

5) Annual prepay framing (compare monthly vs annual anchor). Hypothesis: adding an explicit annual savings anchor increases ARR per signup even if conversion rate shifts slightly. Sample: new signups on paywall, variants with and without annual price framed as X/month billed annually. Metric: percent choosing annual, ARPU (first 12 months projected), cash flow uplift. Implementation: show different price cards; ensure order creation uses the correct billing cadence.

6) Feature gating vs feature trial. Hypothesis: gating a key microfeature behind a paywall but offering a short time-limited trial will increase paid conversion and ARPU among engaged users. Implementation: grant a 3–7 day trial on paywall to the test cohort only and measure conversion-to-paid post-trial.