Half your installs no longer start in the app store
Store search used to be the whole game. It now accounts for slightly under half of installs, and the layer that sits above it does not read your keyword field at all.
For about fifteen years, app discovery had two doors. People searched the store, or the store featured you. Optimise the first, hope for the second. The entire discipline of ASO was built on that geometry.
The geometry changed. App store search now drives roughly 48% of installs, down from about 58% two years ago. Social platforms and AI assistants together account for somewhere between 35 and 40% of new app discovery. Store search is still the largest single channel, and it is no longer a majority.
The discovery layer moved upstream
The behaviour is simple to describe and awkward to optimise for. Someone describes a problem to an assistant, gets a shortlist back, and opens the store already knowing what they intend to install. The store did not make the recommendation. It processed a decision made somewhere else.
In early 2026 OpenAI launched Apps in ChatGPT, built on the Model Context Protocol, letting apps be surfaced and used inside conversations. Pilot partners included Booking.com, Canva, Coursera, Expedia, Figma, Spotify and Zillow, with more expected through the year. ChatGPT reaches over 800 million users. The recommendation is not triggered by a keyword. It is triggered by intent expressed in the middle of a conversation.
The stores have responded by building the same thing inward. Google announced Ask Play at I/O 2026, a conversational layer inside listings that answers natural-language questions about an app before installation, generated from the description, the developer's website and aggregate reviews. Apple now surfaces personalised collections with AI-written notes explaining each recommendation.
Two of the three inputs powering Ask Play sit outside the store. Your website and your reviews are now ranking factors for a surface you do not control.
Why this breaks conventional ASO
Keyword fields are a compression format. They exist because a search box needed short strings. Assistants do not consume short strings. They consume prose, from wherever they can find it, and they cite the web far more readily than they cite a store listing.
Which produces an uncomfortable conclusion for a lot of app teams: the highest-leverage ASO asset in 2026 may be a landing page with proper structured data, a genuine presence in the places models trust, and a review corpus that reads like evidence. AppFollow's chief executive put the shift well when he described the winning teams as those treating store description, reviews and website as one source of truth that any model can quote, rather than three separate marketing assets.
The store did not stop mattering
It changed job. Discovery moved upstream; the store became the verification step. Someone arrives with a name already in mind and checks the screenshots, the rating, the recency of reviews, and whether the app looks maintained. Conversion optimisation on the listing matters more now, not less, because the traffic reaching it is warmer and more decisive.
Apple's ranking has also shifted toward behavioural signals. Apps with stronger retention, more frequent sessions and deeper engagement rank better on competitive terms even when keyword optimisation is identical to a competitor's. Rating recency and metadata freshness carry more weight than they did.
That last shift deserves a note of appreciation, because it points somewhere healthy. When retention becomes a ranking input, the fastest route to visibility runs through building something people keep using. Marketing and product stop being separable, which is inconvenient for the org chart and good for everyone else.
What we would do this quarter
Ask the assistants your buyers use to recommend an app for the job yours does, and write down what comes back. That takes an afternoon and tells you more than a month of keyword research. Then work backwards: if you are not named, find what the models are citing and go be present there.
Keep optimising store search. It is still the largest single channel and will be for a while. Just stop treating it as the front door, because for half your users it is now the hallway.