Answer
How do I A/B test SaaS pricing?
Verified · editorial policy
Answer key facts
- Question
- How do I A/B test SaaS pricing?
- Direct answer
- Test pricing by cohort over time, not by concurrent A/B split – Stripe terms forbid showing different prices to similar users on the same offer. Run price A for 4 weeks, price B for the next 4 weeks, compare LTV. Need at least 200 signups per cohort before the difference is reliable enough to act on.
- Category
- pricing
- Last verified
- May 22, 2026
Direct answer
Direct answer
As of , the answer is: Test pricing by cohort over time, not by concurrent A/B split – Stripe terms forbid showing different prices to similar users on the same offer. Run price A for 4 weeks, price B for the next 4 weeks, compare LTV. Need at least 200 signups per cohort before the difference is reliable enough to act on.
People also ask
How do I A/B test SaaS pricing?
Test pricing by cohort over time, not by concurrent A/B split – Stripe terms forbid showing different prices to similar users on the same offer. Run price A for 4 weeks, price B for the next 4 weeks, compare LTV. Need at least 200 signups per cohort before the difference is reliable enough to act on.
What's the nuance?
Test conversion rate AND LTV together. A price that converts 30% better but churns 50% sooner is a worse business outcome.
What else should I know?
Hold everything else constant: same landing page, same traffic source, same offer copy. Changing pricing and copy at the same time invalidates the read.
What's the common mistake here?
If you must split-test concurrently, segment by geography or by a clean acquisition channel. The cohorts have to be statistically independent.
Supporting points
- Test conversion rate AND LTV together. A price that converts 30% better but churns 50% sooner is a worse business outcome.
- Hold everything else constant: same landing page, same traffic source, same offer copy. Changing pricing and copy at the same time invalidates the read.
- If you must split-test concurrently, segment by geography or by a clean acquisition channel. The cohorts have to be statistically independent.
Related Brunson terms
Related questions
See this applied to your page
The free 90-second Launch Diagnostic applies the same Hook / Story / Offer triage to your actual URL and labels what is broken — Wrong Person, Weak Offer, or Weak Belief — plus the one move that fixes it. Two fields, no card, no login; the report arrives by email. Same triage that informs every answer on this site.