Comparing screenshot messages by audience need

Compare screenshot messages by asking which accurate account of the app helps a defined audience recognise a task. Test coherent alternatives within an eligible visitor group. A stronger store response does not show why people responded, nor prove they understood the message as intended.

Turn a need into a comparison

Suppose an appointment app serves people finding a new time and people managing an existing booking. One screenshot sequence could lead with available times; another could lead with booking status. First decide whether both messages are relevant to the same visitors. If they serve different groups, a broad experiment mixes two audience questions.

For each sequence, specify the need, opening image, visible action, resulting screen and any access condition. A booking-management message should not imply instant access if sign-in is required.

Apple advises using images from the app interface. The opening image should make sense alone.

Changing several screenshots can be necessary to express one message. Record every changed image and caption. The result then concerns the whole sequence, not an individual word or frame.

Keep audience targeting separate from random assignment

A listing experiment compares variants among visitors eligible for that experiment. It does not identify each visitor's reason for arriving. Apple Product Page Optimization selects localisations and treatment traffic for the default product page. A language is not a task-based audience segment.

Audience-specific pages solve another part of the problem. Apple custom product pages can show different screenshots through unique URLs or assigned search keywords. Google Play custom store listings are also described as a way to target specific user segments.

Google Play documents A/B tests on store listings. Comparing the performance of separately targeted audiences is still different from randomly comparing screenshot variants within one eligible group.

For an Australian campaign, confirm the market actually served and whether the depicted task is available there. Australian English wording cannot make an unavailable feature accurate.

Interpret the message result

Choose the store outcome before launch. Apple reports estimated conversion to download or pre-order for Product Page Optimization treatments.

Keep the audience and uncertainty with the result. Apple's reported conversion rate is the percentage of viewers who download or pre-order; it does not establish whether visitors understood the message or completed the task after installation.

If one sequence performs better, describe the sequence and eligible group that were compared. Then decide whether that message fits the current listing or whether a separately targeted page would better serve a different audience. Avoid extending the result to visitors the experiment did not include.