App Store Optimization in 2026: What Actually Moves Installs

App Store Optimization used to mean keywords in your title and a prayer. In 2026, the store listing is better understood as the landing page for every install you will ever get — paid or organic — and the biggest wins come from treating it that way. Ranking still matters; conversion compounds harder. This is the Exp(G) audit order: conversion assets, message-matched pages, experiments, ratings operations, then keywords — with the feedback loop into paid UA that makes every channel cheaper.

If your listing has not changed in six months while your ads have, you are running a museum at the end of a performance funnel.

Conversion rate beats keyword rank

Both Apple and Google increasingly reward listings that turn impressions into installs; conversion feeds ranking, which feeds impressions. A screenshot redesign that lifts page conversion compounds twice — more installs from existing traffic, then better visibility on top. That is why we start every ASO engagement with conversion assets, not keyword spreadsheets.

Apple's own product-page tooling assumes you will iterate: Custom Product Pages for message matching and Product Page Optimization for testing treatments on the default page. Google's equivalent volume lever is Play Console store listing experiments, which A/B tests graphics and text against install and early retention outcomes. Use the platforms' own experiment surfaces before you buy another rank tracker seat — then use trackers to see whether conversion wins also moved visibility (tooling compared in our ASO tools guide).

Your first two screenshots do the work of the whole page

Most visitors never scroll. The first screenshot pair needs to communicate the core value proposition and the emotional payoff in under three seconds — big claim text, real UI, and a reason to care. Feature lists come later, if at all.

Practical rules we use in Exp(G) reviews: one idea per frame; UI large enough to read on a phone in a store search result; social proof as a supporting line, not a sticker salad; and localisation that rebuilds claims for the market rather than translating English word-for-word. Preview videos help when the product is motion-led (editors, games, creators); they hurt when they delay the value prop past the three-second mark.

A useful teardown: screenshot your page next to the top three competitors in your main geo and ask a non-teammate which product they understand first. If yours loses the hallway test, no keyword strategy will save the month.

Custom Product Pages and custom listings: match the ad

Generic pages tax paid traffic. Apple CPPs (up to 35) and Google custom store listings let you align screenshots and promos to the angle that earned the tap. Exp(G) experience: when a TikTok ad about sleep tracking lands on a sleep-focused page instead of a generic one, tap-to-install conversion often jumps meaningfully — pure CPI reduction with no bid change.

Map pages to value propositions, not to channel names. Build one page per job-to-be-done your ads actually sell, route creatives accordingly, and keep the default page as the broad organic catch-all. Full playbook: Apple Custom Product Pages guide. Pair with Apple Search Ads ad variations for the cleanest read on which themes convert.

Run experiments like a growth team, not a design team

On Play, follow Google's own experiment hygiene: test a single asset class at a time, run at least a week to cover weekday/weekend mix, and prefer icons, video, and screenshots for impact. On iOS, use Product Page Optimization for default-page treatments and reserve CPPs for message-matched paid traffic — different jobs, different metrics.

Pre-commit success criteria the way you would for ads: conversion rate at comparable traffic quality, plus a post-install quality check in your MMP so a prettier icon does not buy worse users. Adjust cohort cuts by campaign or CPP ID are enough to catch that failure mode. Tie results back to unit economics with the ROAS calculator when a lift changes your affordable CPI.

Keep an experiment backlog with one hypothesis per card. "Make it prettier" is not a hypothesis; "lead with the streak payoff in screenshot one because ads with streak hooks outperform" is.

Ratings velocity matters more than the average

A 4.7 with fresh, specific reviews beats a 4.8 that has gone quiet. Build the in-app prompt around a genuine success moment — after a win, a streak, a completed workflow — and instrument prompt show rate, response rate, and resulting star mix. Respond to negative reviews quickly; both stores surface developer responsiveness, and public replies are also useful evidence in an AI-answer era.

Do not tip-farm. Ask at moments of demonstrated value, make "not now" easy, and fix the product issues reviews keep repeating — ASO cannot outrun a broken onboarding. Exp(G) experience: the fastest ratings wins usually come from prompt timing, not from rewriting the prompt copy for the twentieth time.

Keywords still matter — in one specific place

For discovery, the title and subtitle (Apple) and title and short description (Google) carry most of the weight; Apple's App Store search guidance still starts from relevance and clear naming. Pick the two or three phrases your best users would actually type, and resist high-volume generics you will never own.

Work keywords as a release habit, not a one-off stuffing exercise: mine competitor gaps (see rank tracking), ship metadata changes with enough runway to measure, and never swap title terms in the same week you redesign screenshots — you will not know what moved. Keyword field odds and ends still help on Apple; they do not compensate for a weak first screenshot.

The paid ↔ organic loop

Paid ads lift branded search and store traffic; strong store conversion makes every paid install cheaper; better conversion and ratings feed rank, which feeds organic. Teams that silo "UA" and "ASO" leave that loop broken. Operationally: every winning ad angle earns a CPP or listing hypothesis; every listing winner informs the next creative batch; blended CAC and payback — not channel vanity — decide budget (paid vs organic framework).

When CPI creeps, ask whether the auction got harder or the page got worse. Often it is the page. Model what a conversion lift does to affordable CPI in the Payback Engine before you approve another bid increase.

Failure modes

• Keyword obsessions while screenshots still look like a feature dump.
• One generic page for six ad angles.
• Experiments that change five variables at once.
• Chasing category volume terms you cannot convert.
• Treating ASO as a project with an end date.

ASO is a monthly loop of asset tests, page variants, and review operations that compounds alongside your paid engine. It is a standard workstream inside our growth engagements. If you want a conversion-first audit — screenshots, CPP architecture, experiment backlog — book a call.

Localisation is not a translation ticket. Store conversion is market-specific. A screenshot claim that wins in the US can underperform in Germany or Japan because the emotional payoff or proof style differs. Prioritise localisation by revenue, not by pride: your top paid geos deserve rebuilt first frames, not subtitled masters. Play's localised listing experiments exist for this reason — use them per locale when volume supports a clean read. Exp(G) experience: the highest-ROI localisation work is usually screenshot one and two plus the subtitle or short description, not a perfect full-description translation.

A 90-day ASO operating cadence

Month one: baseline CVR by source (Search Ads, paid social, organic), screenshot redesign for the default page, and CPP architecture for the top three ad angles. Month two: run Play listing experiments and iOS product-page tests on the riskiest hypothesis; stand up ratings prompt instrumentation. Month three: keyword gap sprint, promote winning CPP elements into the default page, and feed creative with whatever claims the store proved.

Track each milestone against tap-to-install conversion and post-install quality, not vanity impressions. If a test wins CVR but worsens D7 retention in the MMP, roll it back — a prettier page that buys worse users is not ASO progress. Write the cadence on a calendar. ASO that only happens when "we have a release" forever loses to teams that ship metadata and assets on a growth clock.

Tooling that supports conversion-first ASO

Consoles alone can run this playbook, but AppTweak metadata timelines make competitor tests visible, while Sensor Tower and AppMagic size category movement when organic softens for everyone. AppFollow keeps rating velocity operational. Score screenshot headlines with the hook analyzer the same way you score ad hooks. Translate conversion lifts into budget decisions with the Payback Engine, ROAS calculator, and benchmarks tool — a nicer page that attracts worse payers is not a win.

Sources & further reading

• Apple: Custom Product Pages
• Apple: Product Page Optimization
• Apple Help: Overview of Product Page Optimization
• Apple: App Store search
• Google Play Console: Store listing experiments
• Apple Search Ads: searchads.apple.com
• Adjust Help: Cohorts
• AppsFlyer Help: Create A/B tests