How do I set up split testing for pricing pages?

To set up split testing for pricing pages, treat it like a controlled experiment with a clear business goal.

1) Define the goal

Pick one primary metric, such as:

• Trial starts
• Checkout starts
• Paid conversions
• Revenue per visitor
• Qualified lead submissions

For pricing pages, “more signups” is not always better than “more revenue,” so choose the metric that matches your business model.

2) Form a hypothesis

Examples:

• “Removing the yearly discount from the top of the page will increase upgrades.”
• “Showing a highlighted recommended plan will improve conversion.”
• “Switching from monthly-first to annual-first pricing will increase annual plan selection.”

A good hypothesis says:

• what you’re changing
• why you expect it to work
• what metric should improve

3) Decide what to test

Common pricing page variables:

• Price points
• Monthly vs annual default
• Plan names
• Feature comparisons
• CTA copy
• Social proof
• “Most popular” badges
• Page layout
• Number of plans shown
• Currency display
• Free trial vs demo vs direct purchase framing

Test one major change at a time if you want clean results. If you test multiple things at once, use multivariate testing only if you have enough traffic.

4) Set up variants

Create:

• Control: current pricing page
• Variant A: one changed version

Examples:

• Control: 3 plans, annual default
• Variant A: 3 plans, monthly default
• Control: no highlighted plan
• Variant A: recommended plan highlighted

5) Use proper traffic split

Usually:

• 50/50 for A/B testing
• Random assignment per user
• Keep users sticky to one variant so they don’t see different versions on repeat visits

If you have multiple variants, split traffic evenly or use weighted allocation.

6) Track the right events

At minimum track:

• Page view
• CTA clicks
• Checkout initiation
• Purchase completion
• Revenue
• Plan selected
• Refunds or cancellations if relevant

Also track segmentation data:

• Device
• New vs returning users
• Traffic source
• Geo
• Industry or user type, if applicable

7) Make sure the experiment is statistically sound

Avoid ending tests too early. Watch for:

• Enough sample size
• Enough conversions
• Proper test duration
• Day-of-week effects

A pricing page often needs longer testing because conversion volume can be low and purchase decisions can take time.

8) Run one test at a time if possible

If you’re changing pricing pages, avoid overlapping tests that affect the same funnel unless you can isolate them. Otherwise, results get muddy.

9) Analyze more than just conversion rate

Look at:

• Conversion rate
• Average revenue per visitor
• Annual vs monthly mix
• Customer lifetime value, if available
• Refund/cancellation rates
• Down-funnel quality

A version that increases clicks but lowers revenue may not be a win.

10) Roll out the winner and validate

When you find a winner:

• Roll it out gradually
• Confirm the lift holds over time
• Re-test if seasonality or traffic mix changes

Recommended tools

Depending on your stack:

• Optimizely
• VWO
• Google Optimize is discontinued, so use alternatives
• AB Tasty
• Adobe Target
• LaunchDarkly experiments
• GrowthBook
• Statsig
• PostHog experiments

Simple test example

Goal: increase paid signups
Hypothesis: simplifying plan names will reduce confusion and improve conversions
Control: Basic / Pro / Enterprise
Variant: Starter / Growth / Scale
Success metric: checkout completion rate

If you want, I can also help you design a specific pricing page experiment plan, including sample hypotheses and event tracking.