Stop arguing about screenshots and let the traffic decide.
Product Page Optimization finds a better default page. Custom Product Pages give each campaign its own. Both need screenshot variants, and producing those variants is the part that usually stalls the whole programme.
The first screenshot. It is seen by everyone who reaches the page, so a change there has the largest possible sample and the largest possible effect. Test the second screenshot next, then caption angle across the set, then panel order.
Change one thing per test. A variant that changes the first frame, the captions and the palette at once will tell you that something worked and nothing about what.
Custom Product Pages are for intent, not variety
A CPP earns its keep when the audience arriving is genuinely different — not just from a different ad. Useful splits:
By feature intent. Someone who searched for the one thing you do gets a page leading with that thing.
By campaign. The creative that got the tap is echoed in the first screenshot, so the page does not feel like a different product.
By audience. A page for the person evaluating for their team, versus the person evaluating for themselves.
Making four near-identical pages with the captions reworded is not a CPP strategy, it is four times the maintenance.
Why this stalls
The strategy is not the hard part. The hard part is that each variant needs its own screenshot set, at every size, and if producing a set is a multi-day design job then a four-variant test is a month of work nobody schedules.
That is the constraint we build against. A set is a document: fork it, change the one frame under test, re-render that panel rather than the whole set. Variants stay visually consistent with the original because they came from it, which also keeps the test honest — you are testing the change, not an incidental redesign.
Questions
What is the difference between PPO and CPP?
Product Page Optimization tests variants of your default page against each other to find a better default. Custom Product Pages are additional, permanent pages with their own URL that you send specific traffic to. PPO is an experiment; CPP is a destination.
Which should I do first?
PPO, and test the first screenshot. It is the highest-traffic element on your highest-traffic page, so it is where a win is both largest and fastest to detect.
How many Custom Product Pages can I have?
Apple allows a large number — far more than most teams will ever fill. The practical limit is how many distinct audiences you can actually write different screenshots for, which is usually three or four.
Does Google Play have an equivalent?
Yes. Store Listing Experiments cover the PPO role, and custom store listings cover the CPP role, targetable by country, install state or ad campaign.
Do I need a separate screenshot set per variant?
Per variant you are testing, yes — that is the point. What you should not need is to rebuild the design each time; changing a caption or reordering panels should be an edit, not a new project.