Pricing strategy
App Store price testing: how to A/B test prices by country
You can’t show two prices to two halves of one storefront, but you can change the price in a few countries, leave the rest alone, and compare. Done carefully, that country split answers the only pricing question that matters: does the new price earn more?
Can you A/B test prices on the App Store?
Not directly. App Store Connect has no built-in price A/B test: everyone in a storefront sees the same price, and product page optimization tests icons, screenshots and app previews, not prices. The practical method is a country split: change the price in a few countries, keep the rest as a comparison group, and compare per-day sales and proceeds before and after.
What Apple offers, and why none of it is a price test
- Product page optimization. Apple’s overview describes testing up to three alternative app icons, screenshots and previews against your original page, measured by conversion rate. It tests the page, not what the product costs.
- Temporary price changes. You can schedule a price change for the app with a start and end date in selected countries, for up to a year. Useful for running a test window that reverts by itself — but it is a time-boxed change, not a split, so it still needs a comparison group to be read.
- Introductory, promotional and offer-code prices for subscriptions. These discount specific customers or periods; they tell you how an offer converts, not what the regular price should be.
- Paywall experiments inside your app. Showing different products to different users tests which product people choose; it does not tell you what the same product should cost in a storefront.
Because the price is set per storefront and the storefront follows the customer’s Apple Account country, the country is the natural unit of a price test. You lose true randomization, so the design has to make up for it with a comparison group.
The country-split method
A plain before/after reading — “we cut the price and sales went up 30 %” — is unreliable on its own. Seasons, paydays, a featuring, an update, a competitor’s sale or a currency move all change sales at the same time as your price. The country split separates them:
- Test group: the few countries where the price changes.
- Comparison group (control): every other country, where the price stays the same. Whatever moved sales there — the season, an update — most likely moved the test countries too.
- Effect of the price: the test group’s change with the comparison group’s change divided out. As a formula: (1 + test change) ÷ (1 + comparison change) − 1. If the test group is up 21 % and the comparison group up 10 %, the price itself is worth about +10 %.
Measure both groups over the same days: a window before the change and an equally long window after it, compared as per-day averages so a few missing report days don’t skew the result.
Choosing the test countries
- Enough volume. You need dozens of purchases per window in the test group, not a handful. A count of n events varies randomly by about √n, so 100 purchases carry roughly ±10 % noise and 25 purchases ±20 %.
- A reason to test there. Countries with many downloads and few purchases are natural candidates for a lower price; countries that buy far more often than average, for a higher one. See many downloads, few purchases by country for how to find them.
- Similar countries in both groups. If you test Brazil and Mexico, countries such as Chile or Colombia left in the comparison group make it a fairer yardstick.
- Not your biggest market. Keep the country that carries most of your revenue in the comparison group; the test should risk little.
- A handful, not half the world. Two to ten countries is typical. If nearly every country changes, there is nothing left to compare against.
- Nothing else changing there. Avoid countries where you are running local ads, have a localized launch coming, or expect a featuring during the test.
How long to run it
Run whole weeks, because sales differ by weekday: 14 days is a sensible minimum, 28 days for small apps or expensive products. The before window should be just as long and should end the day before the new price starts. Avoid windows that straddle big local holidays or shopping events in only one of the groups, a new app version that changes the paywall, or a sale.
Plan for the reporting lag too. Apple makes daily Sales and Trends reports available the following day (report availability), so the last days of a test arrive after it ends. Decide the length up front and don’t stop early because the first week looks good — early numbers are the noisiest.
What to measure
All of it comes from the daily Summary Sales report, filtered to the product you are testing and grouped by Country Code:
| Metric | What it tells you | From the report |
|---|---|---|
| Units per day | Whether more people bought | Units (refunds are negative) |
| Proceeds per day | Whether you earned more — the number that decides | Units × Developer Proceeds (per unit), in each country’s proceeds currency |
| Proceeds per download | Earnings with traffic swings taken out | Proceeds ÷ first-time app downloads (1, 1F, 1T, F1 rows) |
| Average price paid | That the new price is actually live | Customer Price on the product’s rows |
| Revenue share of the test group | How much of the business the test puts at risk | Test group proceeds ÷ all proceeds |
Compare proceeds in each country’s own proceeds currency, or convert both windows at one fixed rate. Apple’s USD figures in Sales and Trends are estimates based on a rolling average of the previous month’s exchange rates, so a currency move between windows can look like a pricing effect. You can pull the same reports through the API with `GET /v1/salesReports` and your vendor number.
Reading the results: a worked example
A free app sells a $4.99 “Pro” unlock. The developer cuts the price by 30 % in Brazil, Mexico, Colombia and Indonesia, and leaves every other country alone. Fourteen days before against fourteen days after, per-day averages, proceeds converted at one fixed rate:
| Group | Units per day | Change | Proceeds per day | Change |
|---|---|---|---|---|
| Test group (4 countries, −30 %) | 6.0 → 10.5 | +75 % | $17.40 → $21.10 | +21 % |
| Comparison group (all other countries) | 40.0 → 44.0 | +10 % | $142.00 → $156.20 | +10 % |
| Effect of the price | +59 % | +10 % |
Everything sold 10 % more in the second fortnight, test countries or not, so most of the apparent +21 % in proceeds was the season. With that divided out — 1.21 ÷ 1.10 − 1 — the lower price is worth about +10 % in proceeds and +59 % in units. Proceeds per unit fell from $2.90 to $2.01, in line with a 30 % cut snapped to Apple’s price points. Downloads in the test group barely moved (310 → 320 a day), so proceeds per download rose by about 18 %: the gain comes from more of the same visitors buying.
Is +10 % real? The test group sold 84 units before and 147 after; with counts that small, random variation alone can move the ratio by more than ten points, so the result is promising but not conclusive. The honest next step is another 14 days. Had the effect on proceeds been zero or negative, the extra units would not justify the cut and the right move would be to roll back.
Running a price test step by step
- Write down the question and the decision rule first: “If adjusted proceeds per day rise by at least 5 % over 28 days, keep the price in these countries and their tier.”
- Pick one product — the app price, one in-app purchase or one subscription — and two to ten test countries with enough sales.
- Choose the change: −20 % to −40 % for countries that download but don’t buy, +10 % to +30 % where buyers pay unusually often. Work out the break-even: at 70 % of the price you need about 1.43× the purchases to earn the same.
- Note today’s price in every test country so you can restore it exactly.
- Apply the new prices in the test countries only. In App Store Connect, set prices for those countries; for the app price, a temporary change with an end date reverts on its own.
- Freeze everything else for the test window: no paywall changes, no new screenshots, no sales.
- When the window plus the reporting lag has passed, compute per-day units and proceeds for both groups, before and after, and the adjusted effect.
- Keep the price — and extend it to similar countries through a regional tier — or roll back.
Pitfalls
- No comparison group. Before/after alone credits the price with whatever the season did.
- Judging on units. A cut almost always sells more units; only proceeds tell you whether it paid.
- Stopping early. The first few days are the noisiest and the most tempting to read.
- Apple’s automatic adjustments. On storefronts where Apple generates prices from your base, Apple may adjust them for taxes or exchange rates; once you set prices yourself in a country, Apple stops adjusting them there (schedule price changes). Remember that after the test.
- Subscribers already paying. For auto-renewable subscriptions, a decrease applies to existing subscribers automatically; for an increase you choose whether existing subscribers keep their price, and some changes need their consent (subscription pricing). Measure new subscriptions only (
Subscription= new), and for an increase test keep existing subscribers on their price. - One scheduled subscription change at a time. Apple allows one future price change per country per billing plan type, and scheduling another overwrites it, so you cannot queue the test and the rollback together for a subscription.
Keeping the price or rolling back
If the price wins, don’t stop at the test countries: move them and the countries like them into the matching regional tier, so in-app purchases and subscriptions follow the same logic. If it loses, restore the exact prices you noted. For the app price a temporary change ends by itself; for in-app purchases and subscriptions, schedule the old price back. Rolling back a subscription cut is a price increase, with Apple’s notice rules — another reason to test subscription decreases in a few countries first.
Price experiments with RepriceCat
RepriceCat’s Experiments do the country split for you. On any product, choose Run a price experiment, pick the countries and the change, and the other countries — including your base country — become the comparison group. The new prices go through the same review as any change, so you see each country’s new, snapped price before anything is sent to Apple. Results come from Apple’s daily sales reports, so a vendor number is needed.
- Up to 40 test countries; a change from −40 % to +30 %; measured over 7, 14 or 28 days.
- Units in the changed countries and in the countries left alone, and the effect of the price in units and proceeds, as per-day averages against the same number of days before the change.
- Keep the new prices, or roll back to the prices you had.
- The Sales hints “many downloads, few purchases” and “buys N× more often” link straight to the app’s prices, where the experiment starts; every past change also has a before/after impact view under Activity.
Learn how test results feed back into regional pricing tiers, or start the free month and run your first experiment this week.
Frequently asked
Can you A/B test prices on the App Store?
Not within one storefront: everyone in a country sees the same price and Apple offers no price split test. The working method is to change the price in a few countries, keep the rest as a comparison group, and compare per-day sales before and after.
Can product page optimization test prices?
No. Apple’s product page optimization tests alternative icons, screenshots and app previews against your original product page. The price is the same for every visitor in a storefront.
How long should an App Store price test run?
At least 14 days, in whole weeks, and 28 days for small apps or expensive products. Compare against an equally long window before the change, and allow a day or two for Apple’s daily reports to arrive.
How many countries should be in the test group?
Usually two to ten countries with enough purchases to read, while most of your revenue stays in the comparison group. If almost every country changes, nothing is left to compare against.
What should I measure in a price test?
Proceeds per day decide; units per day and proceeds per download explain why. Adjust the test group’s change by the comparison group’s change to remove seasonal effects.
Can I test a subscription price this way?
Yes, but a price decrease also applies to existing subscribers, and an increase can require notice or consent. Measure new subscriptions only and keep existing subscribers on their price when testing an increase.
Try it
Reprice every territory from one base price
Tiers, price-point snapping, one review table, one confirmation — through the App Store Connect API. First month free, then $1 per app per month.