[Webinar recap] What gaming apps get wrong (and right) in the app stores

Episode two of ASO Under Review went to gaming, with AppFollow and Yodel Mobile auditing live listings. Five titles came up: House Flipper, Idle Pirate Tycoon, MovieStarPlanet 2, CyberStrike Manager Idle, and A Maze Go. Different genres, different levels of maturity, and a surprising amount of overlap in what's going wrong.

What follows pulls the thinking out of that session and lays it out on its own.

Gaming players decide faster than any other category. Quicker to install or scroll past than anyone, which means your icon, your first few screenshots, and your title are doing nearly all of the selling, long before anyone reads a word of your description.

You can watch the full recording here. We highly recommend watching it alongside this article if you're on a tight schedule - just jump forward to relevant sections for additional commentary and imagery. 

Rating volume does two jobs

House Flipper has a lot of ratings, and that count works twice. Players read it the way they read a restaurant's review count, before they read a single review. The stores read it as a ranking signal, and a significant one, which matters more each year as the categories get more crowded.

The sentiment holds up too. Over 80% of its iOS reviews are five stars, so the gameplay lands and the experience is solid underneath.

Idle Pirate Tycoon sits at 4.7 on iOS and 4.5 on Android, and Kolibri Games earns part of that by replying to every single review. Not the featured ones. Not the loud ones, but every one of them, negatives and all.

MovieStarPlanet 2 has the opposite shape, with more ratings on iOS and a better average on Android. Normal enough, since the two stores rarely produce the same numbers. The fix for volume is placement. Pull your engagement analytics, find the moments where players are happiest, and check where your review prompt is in that flow. Most fire somewhere convenient for the developer rather than somewhere good for the player.

iOS visibility is sliding while the metadata sits still

House Flipper's iOS visibility fell around 45% over the past year, and its top 10 keyword coverage went with it. The losses weren't spread evenly. Building games and game simulator dropped from the top 5 to somewhere around 15, and those are core gameplay terms, so that one gets felt in installs.

Algorithm updates are arriving more often and hitting harder. You can't opt out of them. What you can do is stop giving the store reasons to treat your listing as stale, because level of optimization is something it rewards, and House Flipper's iOS title and subtitle haven't been touched in a long time. Best practice is roughly every four weeks. In a category where competitors refresh monthly, a year of silence reads as an abandoned listing.

The recovery move is narrow. Pick one keyword you want back, makeover game for example, put it in the title, let the index refresh, then check the rankings a few weeks later.

Android, meanwhile, is doing fine. Whatever that team is doing over there is worth copying over to iOS.

Brand-only titles waste the strongest field

MovieStarPlanet 2 uses its title for the brand and nothing else. Common enough. It's also the field with the heaviest algorithmic weighting, so leaving it at brand means every generic term has to be won somewhere weaker.

You can see the cost in the rankings. Weak discoverability on social games, avatar, dress up, all of which describe the game precisely.

Start on mid and low-volume terms you can realistically hold, get conversion working there with creative tests so you keep the position, then take on social game and avatar game once you have some authority behind you. It's slow. It also widens the top of the funnel, which feeds the ratings problem from the section above.

A Maze Go is the blunter version. The game is about tapping arrows to clear a maze. The word arrow appears in neither the title nor the short description, and the app doesn't show up in the top four organic results for arrow puzzle or arrow puzzle game. That's about an afternoon of work.

House Flipper has a milder case of the same thing. Its title carries builder, when the product is a house makeover game. House would probably convert better and would separate it from the competitors currently sitting next to it.

Worth remembering on Android: search results don't show screenshots at all. At that stage, the title and the icon are the entire listing.

Screenshots have to work at thumbnail size

A Maze Go's gallery looks like a set of mazes. At the size a player sees them, the arrows aren't visible, and the arrows are the whole mechanic. It takes a download and a couple of levels before the game explains itself, which is a lot to ask of a player.

Go through your own tutorial and use the core loop as the storyboard. Zoom in on the mechanic even if it stops looking like the real game HUD. On iOS that's a review risk, so submit it and find out for yourself. Sometimes the reviewers are lovely about it. Sometimes less so.

CyberStrike Manager Idle has a density problem instead. The game is busy on device, and the Google Play screenshots stylize the gameplay on top of that, so the core loop gets harder to read rather than easier. Google Play gives you room for that kind of styling. iOS wants you literal. And the Android screenshots run no captions at all, which leaves players to work out the value proposition on their own. Nobody does that.

Kolibri went the other way and got away with it. Idle Pirate Tycoon is a portrait game, and its iOS gallery is portrait gameplay set inside landscape frames with blurred backgrounds, alternating with painted landscape art. The video is landscape too. How that passed Apple review is anyone's guess. It sits outside the guidelines, near enough, and it stands out in a category where everything looks the same. (Test it before you copy it.)

House Flipper's screenshots are fine. What's missing there is a testing program. Color scheme, messaging, which gameplay moment leads, none of it is being tested continuously, and in gaming that's the thing you test most.

The icon is the only creative in search results

CyberStrike's icon is a special forces character, which suits the game. The whole image is dark though, and the orange visor with the streak behind it doesn't rescue it. Put it on a couple of real devices and it reads as a dark smudge. Add dark mode on top of that, which is how most people run their phones now, and the icon dissolves into the interface around it.

The bigger problem is that it looks like several other shooter icons in the same set of search results. Same character type, same palette, same everything.

Kolibri runs different icons on iOS and Google Play. Whether that came from testing or instinct, it fits the difference between the two audiences. House Flipper dates its Android icon with the year, which is a cheap signal of newness and worth testing on iOS too, providing the year is right.

Video, feature graphics, and a thumbnail trick

A Maze Go has no video, and portrait games can now upload shorts as trailers, so there's no reason for that gap.

There's a trick attached to it. When you put a portrait video on your Google Play listing, the feature graphic stops appearing, and YouTube picks a random frame for the thumbnail unless you choose one. Choose one. It shows up as an extra tile in the gallery, so eight screenshots become nine value propositions, and the ninth autoplays into the video. It wins occasionally and fails more often than it wins, so run it as a test rather than a rule.

The feature graphic still matters, because portrait video only shows on the store page and not in search results. Google's decision, not yours.

Also, the first three to five seconds of that video are the whole decision. Keep it intentional.

One thing coming: Apple's product page headers and search results assets launch officially this autumn. You're not late yet. They give you somewhere to put brand and seasonal work, and they'll need testing like everything else.

In-app events are a second store page

MovieStarPlanet 2 spreads its in-app events out rather than clustering them, so there's always an event live for players. Copy that structure.

The content is the thin part. Somewhere between 60 and 80% of in-app event downloads come from the event card itself, so the card is a conversion surface. Short copy is good. Short copy with no value proposition is decoration. Check legibility too, since the text color is native and you don't control it, so the creative underneath has to be built around that. Aim for a small landing page on every card.

CyberStrike uses promotional content heavily and shipped a 3v3 mode recently, which tells returning players the game is alive. House Flipper ships plenty of updates and reflects almost none of them in promotional content or event activity. That's a lot of work nobody hears about.

And you don't need live ops for any of this. Anything seasonal can be spun around your core loop, which is the note that went to Kolibri.

Reply to reviews, just not all of them

Whether it's worth replying to everything depends on your resources.

With limited hands, take featured reviews first, negative ones especially. Conversion improves when a player sees developers responding, and a reply sitting under an angry review takes some of the heat out of it for whoever reads next. Timing is the part people get wrong. A reply five weeks after the complaint adds salt to the wound.

Draw a line at monetization. If ads pay for the game, a player complaining about ads isn't describing something you're going to fix, and answering every one of those teaches nobody anything. The compromise for automation is letting it answer two or three monetization complaints a week, acknowledge the feedback, and leave it there. Same for anything purely subjective.

Starting with automated replies and want the safest bucket? One to three-star reviews carrying positive sentiment. The "awesome game" two-star reviews. Ask what went wrong, nudge gently toward the rating, and a decent share of them come back with a better rating. AppFollow handles that kind of rule, or you can build the agent yourself, but a human still reads what comes out of it.

CyberStrike is the cautionary version. Ratings dropped in August, most likely a technical issue in the latest build or a monetization change, and the replies slowed down over the same period. Those two things compound each other.

What Google Natural Language wants from your description

CyberStrike's long description runs about 900 characters, with the same text on both stores. It reads like a person wrote it, which is worth something. It also contains the word tactical twice and not one other search term worth the space.

PvP has decent search popularity and a difficulty somewhere around 40 to 50, which makes it takeable. PvE, shooter, battle, and simulator all describe the game and none of them are in there. The old rule about repeating a target keyword roughly five times across the description still does its job.

Idle Pirate Tycoon has the reverse problem. Well-written description, packed with idle and tycoon, both of which are already in the title. Spend that space on terms you don't rank for yet.

For A Maze Go, the long description is what tells Google Natural Language which category to score you in, puzzle games being the obvious target. Fix the text, and the category assignment follows from there.

Then there's the question of handing description writing to AI entirely.

It's genuinely useful for semantic clusters, for working out which terms live near your category. Copy that reads naturally is where it gets weaker, and natural is what GNL rewards, so keep a human on the pen.

The argument on the other side is that AI recognizes AI best. Train it on what GNL likes, meaning positive or neutral language, facts, and your category's keyword clusters, and the output gets tagged more accurately. Same logic applies to answer engine optimization, which isn't reading keywords at all. It reads facts and the questions behind them. Being mentioned in an AI answer is the goal, and higher up is better because everyone reads top to bottom.

Either way, whatever the model produces goes through human hands before it goes to the store.

If you only have resource for one thing

Weigh it against your product's last few releases. Shipped a content update or something seasonal? Put the month into screenshots, because an event announcing new content next to screenshots from six months ago is a red flag to players. Nothing on the calendar? Test the icon. Work out whether you need a video, or whether the one you have needs different content in it. Seasonal icons are cheap to test and easy to repeat every year.

There's a counterargument for spending that time on metadata instead, and it's a chicken-and-egg one. Conversion rate work needs traffic to work on, and if your generic visibility is near zero, there's nothing coming through to learn from. Build some presence, then use those users to find out what converts and why.

The tiebreaker is your traffic. Mostly paid? Leave keywords alone and spend the time on icons and creatives. Mostly organic, in a category where organic is how people find things? Then it's keywords, and the answer gets easier.

Four things to look at this week, if you want the short version. 

• Whether your title names the genre
• Whether your screenshots read at thumbnail size, on a device, in dark mode
• Whether your icon looks like every other icon in your search results
• Whether your metadata has been touched in the last four weeks.

Streaming and entertainment is next in the series. Worth sitting in even if you ship games, since there's usually something to steal from another category. Stay tuned for that one.

cta_get_started_purple

FAQ

What is ASO for a mobile game?

App store optimization covers everything that decides whether players find your game and then install it. Metadata and keywords handle discovery. Icon, screenshots, video, ratings, and event cards handle the decision. In gaming the second half carries more weight than in most categories, because players decide in seconds and rarely read anything.

How often should you update your game's store metadata?

Around every four weeks. The stores treat optimization activity as a signal, so listings that sit untouched for months tend to lose ground, especially on iOS where algorithm updates have been landing harder. Change one thing at a time so you can read the result, and give the index time to refresh before judging it.

Do game screenshots need captions?

On Google Play, yes. Gameplay alone asks the player to work out what the game is and why it's worth the download, and most players won't do that work. A caption can carry social proof, a mechanic, or a reason to download the game. Check legibility against your creative, and make sure the mechanic is still visible at thumbnail size.

Should you reply to every review on Google Play?

Only if you have the people for it. Featured and negative reviews come first, and speed matters more than volume, since a reply weeks after the complaint does more harm than good. Automation covers the repetitive cases well, though it's worth capping replies to complaints you can't fix, like ads in an ad-funded game.

Can AI write your app store description?

It can draft one, and it's genuinely good at finding the semantic clusters that Google Natural Language scores you on. Copy that reads naturally is still where humans win, and natural is what GNL scores highest. Use it for keyword and structure work, then edit properly. Nothing goes to the store unread.