Skip to main content
Your app store listing is a conversion funnel: the icon wins the tap, the screenshots and trailer build interest, the description closes it. This playbook tests each element with real mobile players and then validates the complete listing before you spend on user acquisition. Unlike the e-commerce loops, this is a sequential element test — you optimize each listing component in order, then run one combined validation. The loop is designed to be run by an AI agent (Claude, ChatGPT, Cursor, or any client with PickFu MCP / CLI / API access) or step-by-step by a human.

​
When to use this playbook

Use it whenSkip it when
You’re preparing a mobile game / app store listingYou haven’t validated the concept yet — use game concept validation first
You have multiple icon, screenshot, or description options to testYou have nothing to compare (generate options first)
You want player-representative feedback before launchYou need post-launch retention data (that’s analytics, not pre-test)
Audience targeting. Because this tests a mobile game, the polls target mobile gamers (mobileyes) in the core 25–44 age range and report results by age and gender. Adjust the targeting to your game’s actual demographic — a kids’ puzzle game and a hardcore strategy game have very different players. Use list_available_targeting to see options.

​
The sequence

1

Compare icon designs

The icon wins or loses the tap in the store grid. Rank 3–4 icon variations.
2

Test your gameplay trailer

Open-ended feedback on whether the trailer makes players want to download.
3

Rank your screenshots

Rank 4–8 screenshots by download intent to find your strongest ordering.
4

Test your app description

Compare description versions to find the most compelling copy.
5

Validate the full listing

A multi-question survey on the assembled listing confirms it converts before launch.
Sample size and cost. Defaults to 50–100 respondents per element test (ranked tests use 100 for a clearer winner; open-ended uses 50). For a tighter budget, 50 across the board is fine for directional results. The final validation uses 100. See sample size guidance.

​
Run this playbook with an AI agent

Copy prompt for AI

Paste this prompt into Claude, ChatGPT, Cursor, or any AI agent connected to the PickFu MCP server, CLI, or REST API. The agent will run the entire loop on your behalf — creating polls, reading responses, and iterating until a winning variation emerges.
You are running the PickFu "App store listing optimization (ASO)" playbook
end-to-end. Goal: optimize each element of a mobile game's app store listing and
validate the full listing before launch.

Before starting, ask the user for:
- The game and its genre (e.g. "match-3 puzzle", "4X strategy")
- Icon variations (3-4 images)
- Gameplay trailer (video URL or upload)
- Screenshots (4-8 images)
- Description versions (2-3 text variants)
- A screenshot of the full assembled listing (for step 5)
- Target audience (default: mobile gamers 25-44; refine to the game's
  actual demographic if the user knows it)

Substitute [GENRE] with the user's genre in every question. Run in sequence:

1. ICON (100 respondents, ranked, 3-4 options).
   If the user doesn't already have icon variations, GENERATE them: use
   generate_image to create 3-4 distinct icon concepts (vary the hero element
   — character face, key object, color treatment). Character-focused icons
   tend to outperform logos in gaming, so include at least one character
   concept. Then run a ranked poll: "Which game icon makes you MOST want to
   download and play a mobile [GENRE] game? Explain what catches your eye."
   Targeting: mobile gamers, 25-44. Report by age + gender.
   If no icon reaches 35%+ first place, take the strongest elements from the
   top two, generate a new concept combining them, and re-run replacing the
   weakest — an iterate-to-winner sub-loop, same shape as the Amazon main
   image playbook. The winner is your launch icon.

2. TRAILER (50 respondents, open_ended).
   Upload the trailer. Question: "Watch this gameplay trailer for a mobile
   [GENRE] game. Would you download it? What excites you or puts you off?"
   Use the feedback to recut if needed.

3. SCREENSHOTS (100 respondents, ranked, 4-8 options).
   Question: "Which image would make you MOST likely to download and play a
   mobile [GENRE] game? Tell us what about the image makes you want to play."
   The ranking gives you the strongest screenshot order.

4. DESCRIPTION (100 respondents, ranked, 3 options).
   Provide 3 description versions as text options (use ranked, which needs 3+
   options; if you only have 2 versions, use head_to_head instead). Question:
   "Which app description would make you MOST likely to download this mobile
   [GENRE] game? What about the description appeals to you?"

5. VALIDATE FULL LISTING (100 respondents, multi-question).
   Upload a screenshot of the assembled listing as the context image for Q1.
   Q1 (star_rating): "Based on this app store listing, how likely are you (1-5)
   to download this game?"
   Q2 (open_ended): "What's the most appealing thing about this game based on
   the listing?"
   Q3 (open_ended): "Is there anything confusing or that would stop you from
   downloading?"
   Q4 (multi_select): "What type of game does this look like?" with options
   Strategy / Action / Puzzle / Casino / Sports / RPG / Simulation / Other.

   Success target: full listing averages 3.5+ stars on Q1, and Q4 shows players
   correctly identifying the genre.

Final report must include:
- The winning icon, screenshot order, and description
- The full-listing star average and the top appeal + top blocker from Q2/Q3
- Whether players correctly identified the genre (Q4)
- Any element scoring poorly enough to redo before launch

Tools to use:
- save_survey + publish_survey         — create and launch each poll
- get_survey_responses                 — read responses
- generate_image                       — create/iterate icon concepts (step 1)
- list_available_targeting             — confirm targeting codes
- upload_media                         — screenshots, trailer, finished assets
Want to run this manually? The same sequence is available as a one-click template in the PickFu app (Start the ASO playbook) with pre-filled poll URLs.

​
Step-by-step (human operator view)

​
1. Compare icon designs

SettingValue
Poll typeRanked choice
Question”Which game icon makes you MOST want to download and play a mobile [GENRE] game? Explain what catches your eye.”
Options3–4 icon variations
AudienceMobile gamers, 25–44 (report by age + gender)
Sample size100
No icon variations yet? Generate them with generate_image. Create 3–4 concepts varying the hero element (character, key object, color), include at least one character-focused concept (they tend to win in gaming), then iterate to a winner the same way the Amazon main image playbook does — combine the strongest elements of the top two and re-run until one icon clears 35%+ first-place votes.
See an example icon poll →
PickFu ranked poll comparing four mobile RPG game icons with vote share and AI insights

​
2. Test your gameplay trailer

SettingValue
Poll typeOpen-ended
Question”Watch this gameplay trailer for a mobile [GENRE] game. Would you download it? What excites you or puts you off?”
OptionsYour gameplay trailer / video preview
AudienceMobile gamers, 25–44
Sample size50
See an example trailer poll →

​
3. Rank your screenshots

SettingValue
Poll typeRanked choice
Question”Which image would make you MOST likely to download and play a mobile [GENRE] game? Tell us what about the image makes you want to play.”
Options4–8 screenshot options
AudienceMobile gamers, 25–44
Sample size100
See an example screenshot poll →

​
4. Test your app description

SettingValue
Poll typeRanked (3+ versions) or Head-to-head (exactly 2 versions)
Question”Which app description would make you MOST likely to download this mobile [GENRE] game? What about the description appeals to you?”
OptionsYour description versions, as text options
AudienceMobile gamers, 25–44
Sample size50–100
Ranked polls require 3–8 options. If you’re comparing only two description versions, use a head-to-head poll instead — a 2-option ranked poll will be rejected.
See an example description poll →

​
5. Validate the full listing

SettingValue
Poll typeMulti-question survey
Q1Star rating: “Based on this app store listing, how likely are you (1–5) to download this game?”
Q2Open-ended: “What’s the most appealing thing about this game based on the listing?”
Q3Open-ended: “Is there anything confusing or that would stop you from downloading?”
Q4Multi-select: “What type of game does this look like?” (Strategy, Action, Puzzle, Casino, Sports, RPG, Simulation, Other)
ContextUpload a screenshot of the full app store listing as the Q1 context image
AudienceMobile gamers, 25–44
Sample size100
Success criteria: a clear winning icon (35%+ first-place votes), screenshots ranked by download intent, a validated description, and a full listing averaging 3.5+ stars on Q1 — with Q4 confirming players correctly read the genre. See an example full-listing validation → These are the demographics most likely to play each genre. You know your audience best — adjust as needed, and use list_available_targeting for the exact trait codes.
GenreTargetingWhy
Strategymobile gamers, 25–34, 35–44, maleStrategy core demo
Casino / Slotsmobile gamers, 35–44, 45–54, 55–64Casino skews older
Sportsmobile gamers, 18–24, 25–34, maleSports skews young male
Casual / Puzzlemobile gamers, 25–34, 35–44, femaleCasual skews female
RPG / Fantasymobile gamers, 18–24, 25–34RPG skews younger

​
Pro tips

  • Character-focused icons beat logos. Across hundreds of gaming icon tests, icons with a character face or figure out-perform abstract logos and text-only icons. Always include a character concept.
  • Test icons at display size. Upload icons at roughly 60×60px — the size they appear on the store shelf. Performance at small size is what matters, not full-resolution polish.
  • Lead with your strongest screenshot. Position 1 gets 3–5× more views than position 4+, so your #1-ranked screenshot should occupy the first slot.
  • Vibrant colors win on the shelf. Saturated treatments consistently out-perform muted or dark ones in icon testing.
  • Re-test after major updates. Re-run the screenshot and full-listing steps after any major update or seasonal re-skin so the listing reflects the current game.

​
Troubleshooting

Ranked polls need 3–8 options. If you only have two description versions, switch the poll type to head-to-head (exactly 2 options). Add a third version to keep it ranked.
If the multi-select genre answers scatter or point to the wrong category, your icon + screenshots are sending mixed signals. Revisit steps 1 and 3 — the winning assets should make the genre unmistakable at a glance.
Target the demographic that matches your game. A hardcore strategy game tested on casual players gives misleading results. Use list_available_targeting to find the right mobile-gamer and age filters, and report by age + gender to spot segment differences.