What can you A/B test with Product Page Optimization?

Icons, screenshots, app previews and promotional text — up to three treatments against your current page, each receiving a share of traffic, for up to 90 days. What PPO cannot test: title, subtitle, keyword field, description or price. Metadata experiments remain ship-and-measure territory.

Why do PPO icon tests require an app update?

Icon variants must be included as assets in your shipped binary, so testing icons means submitting a build containing the alternates first. Apple requires this so a user who installs from a variant page sees that same icon on their home screen — the store page and the device stay consistent.

How long should a PPO test run for a small app?

Until enough conversions accumulate per treatment, not until impatience wins. Low-traffic apps often need most of the 90-day window for a trustworthy read on modest differences; a few hundred conversions per treatment is a reasonable floor for detecting anything but dramatic wins. Declaring a winner in week one on 40 installs is astrology with a progress bar.

FunnelHound / Guides / A/B testing with PPO

App Store A/B testing with Product Page Optimization: the honest guide

By Emin — indie iOS developer building FunnelHound · Published 16 Sep 2026

Apple will genuinely A/B test your store page: up to three treatments against the current page, real traffic split, up to 90 days. The two catches nobody leads with: PPO can't touch title, subtitle or keywords, and small apps rarely have the traffic to call a winner quickly — patience is a statistical requirement, not a virtue.

What's testable — and what isn't

In scope: icons, screenshots, app previews, promotional text. Out of scope: title, subtitle, keyword field, description, price. That boundary defines PPO's job: it optimizes the conversion stage — what happens after someone lands on (or sees) your page — while the discovery stage stays ship-and-measure per the usual discipline. So run PPO to fix a weak conversion percentile, not a visibility problem: prettier screenshots on a page nobody finds is remodeling a house with no road to it.

The icon exception (in both directions)

Icons are PPO's special case twice over. Logistics: icon variants must ship as assets in your binary, so an icon test starts with an app update containing the alternates — plan a release ahead. Power: the icon is the one PPO element that also appears in search results, so it's the only treatment that can move tap-through, not just page conversion. If you get one well-powered test a quarter, the icon is usually the highest-leverage candidate — it campaigns on every surface your app appears.

Reading results without fooling yourself

Volume first. A few hundred conversions per treatment is a sane floor for detecting modest differences. Small apps may need most of the 90 days — that's the deal.
Resist the week-one peek. Early leads flip constantly at low volume. A 60%-confidence "winner" is a coin with marketing.
Mind the traffic mix. A featuring spike or ad burst mid-test floods treatments with different-intent visitors. Note external events; extend past them.
One hypothesis per treatment. "New screenshots, new order, new captions" winning tells you nothing about why. Change one argument, learn one fact.
Test big swings. Outcome-first caption versus feature list, different first frame (the billboard rules) — not #EB6834 versus #EB6836. Small apps only have power to detect big effects, so only test big ideas.

After the test: confirm in the funnel

Apply the winner, then verify the promoted treatment at full traffic: PPO's split happens inside a period's mix, and the honest final read is your conversion rate before versus after rollout, same weekdays, mix noted. Mark the rollout as a milestone in FunnelHound and let the funnel confirm what the test predicted — trust, but verify, especially when the test squeaked in at the edge of significance.

PPO predicts. Your funnel confirms.

FunnelHound tracks conversion from App Store Connect with the rollout milestone marked — the after-the-test answer, on your iPhone.

Get FunnelHound

Share this on X

Data notes: PPO capabilities, treatment limits and the 90-day window per Apple's App Store Connect documentation as of 2026. Conversion-count guidance is directional statistics, not an Apple rule. PPO availability can vary by page type and storefront — check current docs before planning a quarter around it.