The Reliability Problem
We are seeing a consistent pattern in practitioner reports: apps that rank well for keywords are generating zero impressions. One developer tracking their own app found literally "0" impressions across all ranked keywords in a single day — a signal that something structural has shifted in how wiki:keyword-ranking data reaches ASO tools, or in how the App Store itself weights those positions.
Two theories are circulating. The first: Apple changed or removed algorithmic features starting around October 2025, creating drift between what tools report and what the store actually delivers. The second is more systemic: the proliferation of ASO tools and developer queries has artificially inflated wiki:search-volume signals. If thousands of entities query the same keywords daily through API-driven tools, that volume registers as popularity — but it is not real user demand. The metric becomes a reflection of the ASO industry observing itself rather than user behavior.
Both explanations point to the same practical outcome: keyword position tracking alone is no longer a reliable proxy for visibility or installs. What practitioners thought they were optimizing for — rank on a list — may no longer correspond to what the algorithm actually delivers.
What Changed in the Algorithm
The App Store's ranking model in 2026 operates differently than it did two years ago. Custom Product Pages now participate in organic search as of July 2025, expanding the indexable surface from one listing to up to 70 pages per app. This allows apps to target segmented intents and keyword clusters that a single listing could never cover coherently.
More importantly, retention metrics now carry ranking weight comparable to metadata. Apps with strong Day 1 and Day 7 retention are climbing in search results even without keyword density optimizations. Conversely, apps that generate install velocity through paid campaigns but fail to retain users are losing positions faster than before. The algorithm is reading user behavior as a quality signal, and it outweighs keyword stuffing.
Semantic search has also improved. Exact keyword matches are no longer the only path to indexation. The store now understands intent and synonyms well enough that covering every possible keyword permutation has diminishing returns. What matters more is whether the page and the product align with what users are actually searching for — and whether those users stay.
The Practitioner Gap
Most developers update their store listing once at launch and return only when positions have already fallen. In that window, the algorithm has shifted how it reads the page multiple times. We are tracking cases where teams spend hours polishing icons while their semantic coverage addresses only half of the relevant query space. Or they add new keywords without understanding why wiki:conversion-rate is not improving — confusing ranking factors with conversion factors.
Ranking factors determine which queries surface your app and at what position: title, subtitle, keyword field, user retention, install velocity, rating dynamics. Conversion factors determine whether a user installs after landing on your page: screenshots, icon, description, video, reviews. The former gets you seen. The latter gets you chosen. Mixing the two leads to misallocated effort.
The most common errors we are observing:
- Repeating keywords across title, subtitle, and keyword field. On iOS, this does not add weight — it wastes the 100-character keyword field on redundancy.
- Optimizing for high-frequency keywords without matching user intent. An app can rank first and convert zero installs if the query does not align with what the product delivers.
- Ignoring the first three screenshots visible in search results before a user even clicks through. If those do not communicate value in one second, CTR collapses, and positions follow.
- Not testing visual assets. Teams that run A/B tests every two weeks accumulate 25 learning cycles per year. Those that test quarterly get four.
- Translating text without localizing visuals or intent. In markets with low English proficiency, English screenshots kill conversion regardless of how well the metadata is translated.
What Works Now
The title remains the highest-weighted ranking element. Keywords in the app name receive more algorithmic trust than the same words anywhere else. The standard formula is brand plus one to two high-frequency terms, optimized for character economy: colons instead of dashes, ampersands instead of "and."
The subtitle is 30 characters that influence both ranking and click-through rate because it appears directly in search results. The best subtitles explain value rather than list features. If the core keyword is not in the first half, users on smaller screens will not see it.
The keyword field is 100 characters of semantic coverage invisible to users but indexed by the algorithm. The most expensive mistake here is repeating terms already in the title or subtitle — those 100 characters should expand your query footprint, not reinforce what already works.
Custom Product Pages are now a ranking lever, not just a paid-traffic tool. Each CPP can target a distinct keyword cluster with its own metadata and creative. A fitness app can have one page optimized for weight loss queries and another for prenatal yoga, each with screenshots and messaging tailored to that intent. The algorithm indexes all of them.
Screenshots do not contribute to text indexing, but they influence ranking indirectly through conversion behavior. If the first three screenshots — visible in search before a user clicks — do not clarify the product's value, CTR drops. Low CTR reduces conversion. Low conversion signals to the algorithm that the listing does not match the query, and positions decline.
Ratings below 3.5 stars correlate with reduced visibility. Above 4.5, the algorithm treats high ratings as a persistent quality signal. But it is not just the average — the trajectory matters. Fresh reviews carry more weight than historical ratings. An app that moved from 3.8 to 4.7 recently will outrank one that has held 4.5 static.
The store also indexes review text. If users frequently mention specific features or terms in reviews, that influences search result ranking for related queries. Managing reviews is not just reputation work — it is part of your semantic strategy.
Measurement and Iteration
Keyword position tracking remains useful as a baseline signal, but it is no longer sufficient. What matters is conversion rate from search by query. If you rank well but convert poorly on a keyword, the mismatch is in the listing or the product — not the algorithm.
Competitor monitoring shows where your semantic coverage has gaps. If a competitor climbs on a keyword where you previously ranked, that is the first indicator to revisit priorities.
Visibility score — an aggregate measure across all tracked queries — helps assess overall movement without drilling into every individual keyword. But the metric that closes the loop is retention. If users install and leave, the algorithm will pull your positions regardless of how well you execute metadata.
The fastest audit: check for keyword duplication across title, subtitle, and keyword field. Review what is visible in search results before a user clicks. Compare conversion rates across traffic sources in App Store Connect. Check when you last updated metadata — if it has been more than two months, the competitive and algorithmic landscape has shifted without you.
The Real Problem
Most teams do not lose positions because they chose the wrong keywords. They lose positions because the listing promises one thing and the product delivers another — and the algorithm detects that gap through user behavior faster than the team realizes it exists.
Metadata can be fixed in a day. Product-market fit cannot. The shift toward retention-weighted ranking exposes that misalignment immediately. If your keyword strategy is perfect but your Day 1 retention is 15%, you will not hold positions no matter how well you optimize the page.
The other structural issue is cost. Premium ASO tools with accurate data now run $99+ per year, which many indie developers cannot justify. This is driving a wave of free or low-cost alternatives built by practitioners who only need keyword monitoring and rank tracking. The trade-off is feature depth and data refresh speed, but for developers optimizing one or two apps, the gap is acceptable.
What we are tracking is a moment where the discipline is re-aligning around first principles. The era of mechanical keyword insertion and rank-chasing is closing. What replaces it is harder: building products that retain users, aligning page promises with product delivery, and iterating on creative faster than competitors. The algorithm is not broken — it is just measuring what it should have measured all along.