iOS Product Page Optimization: setting it up and what to test first

Product Page Optimization is Apple's built-in experiment tool. It is free, it uses your real App Store traffic, and most apps have never run one.

This is the iOS-specific companion to A/B testing your screenshots, which covers the statistics. Here we are concerned with the tool and the running order.

What PPO actually does

You create up to three treatments alongside your current product page. Apple splits App Store traffic between them and reports how each performs against the baseline, with an indication of confidence.

What you can vary:

• App icon — with a caveat below
• Screenshots
• App previews

What you cannot vary: name, subtitle, description, keywords. Those belong to the page itself, and changing them is a metadata update rather than a test.

Setting one up

1. In App Store Connect, open your app and go to Product Page Optimization under the Features section.
2. Create a test, name it something you will still understand in three weeks — "Treatment B" tells you nothing later.
3. Choose how many treatments, up to three.
4. Set traffic proportion. This is the share of visitors who see a treatment rather than your baseline.
5. Upload assets for each treatment. Treatments can differ per localization, which matters if you ship several languages.
6. Start it, and then do not touch it.

On traffic proportion: higher gets you to confidence faster and exposes more visitors to a possibly worse page. If your listing has meaningful volume, a smaller share is the safer trade. If it is quiet, you need a larger one or the test will never conclude.

What to test, in order

The order is not arbitrary — it follows audience size, because effect detectability follows audience size.

1. The first screenshot

Everyone who reaches your page sees it. Largest sample, largest possible effect, cheapest change.

Do not test a caption tweak. Test a genuinely different approach: outcome versus feature, UI-forward versus concept-forward, people versus product.

2. The second screenshot

Second-largest audience. Same reasoning.

3. Caption angle across the whole set

Benefit-led against feature-led is a real strategic difference and one of the few whole-set changes worth a treatment.

4. Panel order

Cheap to produce — the assets already exist — and occasionally surprising.

5. The icon

Potentially the largest effect on the list, with two complications: it needs to ship in the binary, and it changes recognition for people who already have your app. Worth doing, worth doing carefully, not worth doing first.

Reading the result

Apple reports improvement against the baseline with a confidence indication. Three disciplines:

Wait for it. Early swings are noise. Decide the stopping rule before you start and hold to it. Checking daily and shipping the day your variant is ahead is how teams ship losing pages with confidence.

One variable per treatment. A treatment that changes screenshots and previews and the icon tells you something moved and nothing about what.

Check per-localization results if you ship several languages. A treatment that wins overall can be losing badly in your second-largest market, and the average hides it.

PPO is not CPP

Easy to conflate, and they solve different problems:

	PPO	CPP
Purpose	Find a better default page	Match a page to an audience
Lifetime	Temporary experiment	Permanent
Traffic	Split from your existing	Sent by a link you control

If you want to serve a different page to a specific campaign, that is a Custom Product Page, not a test.

The real blocker

It is never the tool. It is that three treatments means three full screenshot sets, at every size, and a team that needs a designer for each one simply never starts.

Make the second version cheap and this whole practice becomes routine: fork the set, change the frame under test, re-render that panel. Everything else stays identical, which is also what keeps the test honest.