LinkSpot turns a product gallery into something a shopper can buy from directly: hotspots on the image, badges, captions, expandable annotations. The whole app happens on the storefront. The admin exists to configure it and is otherwise beside the point.

So the set opens on the gallery, and the admin does not appear until the fourth image.

Show the surface the app changes

The rule that decides a set is simple: find the surface the app actually changes, and open there. For an inventory tool it is the admin. For a checkout app it is the checkout. For LinkSpot it is the product photograph, with the app's marks laid over it.

That sounds obvious and is broken constantly, because the person writing the listing has spent months in the admin and forgets that a merchant has not. The admin is where the work happened. It is rarely where the value is.

The feature image does one thing

It shows a gallery with the annotations on, and the words Turn your product gallery into a sales tool. Nothing else. No panel, no chrome, no browser window, no logo on a colour — all four of which are refused or wasted under requirement 4.4.4.

Six images is not many. The feature image is the one that appears in search results and category pages, where the other five never go. Spending it on anything other than the single clearest statement of what the app does is spending your only guaranteed impression on a detail.

Different features, not different angles

The five screenshots take one capability each: annotations, captions, the badges, the editor, and the way it behaves on a phone. Every one is a different feature, view or state, which is what requirement 4.4.5 asks for in those words.

The failure mode here is subtle. An app whose value is visual is very easy to illustrate with six beautiful pictures of the same feature at six angles. They look like a considered set. They are one image submitted six times, and they are refused under the same rule as six crops of a settings page.

One image for the objection

The fifth screenshot is the editor, and it is there for one reason: a merchant looking at a gallery covered in hotspots wants to know how long it takes to place them. An image that answers the obvious objection is worth more than a sixth image that repeats the pitch.

That is usually where a set is won. Five images say what the app does. The sixth should answer the thing the reader is quietly worried about — setup time, theme compatibility, whether it slows the site down — because that is what stops an install.