TL;DR
For most SaaS companies: Use ExperimentHQ for marketing site tests (pricing pages, landing pages) and PostHog or Statsig for in-app experiments. Key tests to run: pricing page layout, trial length, onboarding flow, and upgrade prompts. Avoid Optimizely unless you have $50K+ budget.
Who this is for
- SaaS founders and growth teams
- Product managers optimizing conversion funnels
- B2B SaaS companies with freemium or trial models
Who this is NOT for
- E-commerce (see Shopify guide)
- Enterprise SaaS with dedicated CRO teams (consider VWO/Optimizely)
High-Impact Tests for SaaS
| Test | Impact | What to Test |
|---|---|---|
| Pricing page layout | High | 3 plans vs 4 plans, feature emphasis, annual toggle |
| Trial length | High | 7-day vs 14-day vs 30-day free trial |
| Onboarding flow | High | Guided setup vs self-serve, checklist completion |
| Upgrade prompts | Medium | Timing, messaging, and placement of upgrade CTAs |
| Feature gating | Medium | Which features to gate behind paid plans |
| Social proof | Medium | Customer logos, testimonials, case study placement |
Tool Comparison
| Tool | Price | Trial Tracking | Pricing Tests | Verdict |
|---|---|---|---|---|
| ExperimentHQ | $29/mo | Best Overall | ||
| PostHog | Free - $450/mo | Best with Analytics | ||
| Statsig | Free - Custom | Best for Developers | ||
| LaunchDarkly | $10/seat/mo | — | Best Feature Flags | |
| Optimizely | $50,000+/yr | Enterprise only |
Our Recommendation
For marketing site (pricing, landing pages): Use ExperimentHQ. Visual editor, quick setup, $29/month.
For in-app experiments: Use PostHog (if you want analytics bundled) or Statsig (if you want feature flags).
For pricing experiments: Use server-side testing to avoid showing different prices to the same user.
FAQ
Can I A/B test pricing in SaaS?▼
Yes, but carefully. Test pricing page layout and presentation first (plan names, feature emphasis, annual toggle). Only test actual prices with server-side testing to ensure consistency.
Should I use the same tool for marketing site and app?▼
Not necessarily. Marketing site tests (visual changes) work well with visual editors. In-app experiments often need code-based tools with feature flags. Many teams use two tools.