Annual vs monthly pricing, and how to test which to show first

To test annual vs monthly pricing, split new visitors between two versions of the pricing page, one with annual billing selected by default and one with monthly selected. Keep everything else the same. Judge the test on revenue per visitor at 30 and 90 days, and on projected 12-month revenue per visitor, because the two versions collect money on different schedules.

The trap is cash. An annual-first page collects a year of revenue up front from every annual buyer, so it looks far ahead in the first month even if it wins fewer customers. The honest comparison is what each version is likely to earn over a year. Decide before launch which number picks the winner, and don't switch after you see the first month.

Why the default matters

Many people keep whatever is already selected. In Johnson and Goldstein's study of organ donation, countries where people had to opt out reported much higher consent than countries where people had to opt in. A billing toggle is a smaller decision, but a pre-selected option can pull buyers the same way.

Annual plans also change how long customers stay. RevenueCat's analysis of 10,000+ subscription apps found a median of 28% of annual subscribers still subscribed after a year, against 11% for monthly and 3% for weekly. That is consumer app data. B2B software can look very different, so use your own numbers where you have them.

What each version is likely to change

None of these is certain for your business. That is why you test.

A worked example

This is a made-up example. A SaaS product costs $30 a month or $288 a year (20% off, or $24 a month). Each version gets 6,000 new visitors.

For first-month cash, A collects 144 × $30 + 36 × $288 = $4,320 + $10,368 = $14,688. B collects 73 × $30 + 89 × $288 = $2,190 + $25,632 = $27,822. On this number, B is 89% ahead.

Projected 12-month revenue needs one more input, the number of times a monthly customer pays in their first year. Suppose your data shows 85% of monthly customers renew each month. The expected number of payments in 12 months is 1 + 0.85 + 0.85² + ... + 0.85¹¹, which equals (1 − 0.85¹²) ÷ 0.15, or about 5.7.

• A: 144 × $30 × 5.7 = $24,624 from monthly customers, plus $10,368 from annual, for $34,992.
• B: 73 × $30 × 5.7 = $12,483 from monthly customers, plus $25,632 from annual, for $38,115.

On a 12-month view, B is 8.9% ahead, not 89%. It won fewer buyers and still earns more because annual customers can't churn in months two to twelve. That is a real but modest gain, and on 6,000 visitors per version it may not be distinguishable from noise. Run the numbers through the revenue per visitor calculator before you call it.

Two more checks before you ship B:

• Refunds. If your refund policy lets annual buyers get money back in the first 14 or 30 days, subtract those refunds from B's cash.
• Year two. Annual customers renew or leave all at once at month 12. RevenueCat's renewal data found a median first renewal rate of 27% for annual app subscriptions, against 56% at the first monthly renewal. Keep a holdout group to see how year two plays out.

How long to run it

You need enough visitors to detect the difference in projected revenue, which is usually smaller than it first looks, and enough time to check your retention assumption. A practical plan:

1. Run until you reach the sample size for the lift you care about. The sample size guide and test duration guide explain how.
2. Read revenue per visitor at 30 days, including the first monthly renewal attempts and early refunds.
3. Read again at 90 days and update the 12-month projection with the real renewal rate of the monthly customers in the test, since the default may change who chooses monthly.

How to show the prices

Show the amount people will actually be charged clearly. For apps, Apple's subscription guidance requires the billed amount to be the most prominent pricing element, with any monthly equivalent or savings shown in a smaller, secondary position. On the web you have more freedom, but "$24/mo" in large type with "billed $288 yearly" in small grey text is the kind of display that causes refund requests and chargebacks. Test the display, and track refunds and support tickets alongside revenue.

Auto-renewal rules also apply to annual plans. California's law, as amended from July 2025, requires express consent to the renewal terms and a yearly reminder, among other things, as Cooley's summary explains. This is general information, not legal advice.

When the answer differs by visitor

The best default can differ between groups of visitors. Splits worth checking after the main test ends:

• Team buyers vs individuals. Someone buying for a company may prefer one annual invoice, while an individual may prefer a small monthly charge.
• Countries. An annual price is a large number in some currencies, and card and invoice habits differ.
• People who already use a free plan or trial vs brand new visitors. Someone who already uses the product takes less risk by paying for a year.

Treat these as ideas for the next test, not conclusions. Slicing one test many ways produces false wins, because some slice will look different by chance. If a group looks very different, run a new test for that group alone.

If the main test ends in a draw, keep whichever default matches what most of your customers already choose, and move on to a bigger change, such as the size of the annual discount.

Tests to try, and what to measure

1. Annual selected by default vs monthly selected by default. Measure projected 12-month revenue per visitor, cash per visitor at 30 days and the annual share.
2. A toggle vs both prices shown on every plan card. Measure revenue per visitor and the annual share.
3. Annual discount size: two months free (about 17%) vs 20% vs 25%. Measure projected 12-month revenue per visitor. A bigger discount has to win enough extra annual buyers to pay for itself.
4. Saving shown as a percentage vs a dollar amount ("Save 20%" vs "Save $72"). Measure the annual share and revenue per visitor.
5. Annual shown as the full yearly price vs a monthly equivalent. Measure revenue per visitor, refunds and support tickets.
6. Annual billing offered only on higher tiers. Measure revenue per visitor and plan mix.
7. A "switch to annual" offer after the first or second monthly payment. Measure revenue per customer at 90 days and the share who switch.

Outtest judges tests like these on revenue per visitor pulled from Stripe or another payment tool, and any pricing change waits for the owner's approval before it goes live. For the wider set of pricing experiments, see how to A/B test your pricing.