What AI search is doing to referral traffic, where attention is moving, and why an owned app is the most durable channel a creator or business can build.
For two decades the path to an audience ran through search. Publish something worth finding, earn a ranking, and a steady stream of visitors arrived. That path is closing. AI search now answers the question on the results page, and the click that used to follow never happens.
The durable answer is not a better ranking. It is a channel you own: a place on the home screen, permission to reach people directly, and first-party data that no algorithm can take away. On mobile, that means an app.
We are not arguing this from the outside. We built one for ourselves. ThisWeek.AI went from a 2022 newsletter to a native iOS app, and it now reaches its readers through notifications on a surface we control. This brief lays out the data behind that decision, and the blueprint for making the same move.
AI Overviews are no longer an experiment at the edge of search. They now appear for roughly 13% of all queries, a share that doubled in a single year, on a product with more than two billion monthly users. For a growing slice of searches, the answer arrives before any link does.
The effect on clicks is measurable. Zero-click searches, where the user never leaves the results page, rose from 56% to 69% in a year, according to Similarweb. Pew Research found that people click a result 8% of the time when an AI summary is shown, versus 15% without one, and click a link inside the summary itself only about 1% of the time.
The pages that earned the most traffic are losing the most. Ahrefs reported in February 2026 that top-ranking pages see a 58% lower click-through rate when an AI Overview appears, nearly double the decline measured the year before. Position-one click-through for informational queries fell from about 7.6% to 3.9%, and to roughly 1.6% when an AI Overview triggered.
Publishers feel it most. Chartbeat measured a 33% decline in Google referrals across more than 2,500 news sites in 2025. Search fell from 51% of news traffic in 2023 to 27% by the fourth quarter of 2025. Chegg reported a 49% drop in non-subscriber traffic. Digital Content Next found most of its members down between 1% and 25%, with some off more than 75%.
The legal system has noticed. Penske Media filed suit over AI Overviews, and the European Publishers Council lodged a complaint with regulators. The argument in both is the same: the search engine is now keeping the audience it used to send onward.
If search is one leg giving way, social is the other. The reach a creator builds on a social platform is rented, not owned. It can be changed, throttled, or removed at any time, for reasons the creator never sees and cannot appeal.
2025 made the risk concrete. In January, TikTok went dark in the United States ahead of a federal deadline, and creators who had built their entire audience there lost their line to it overnight. By mid year, Instagram had suspended thousands of accounts, some with little explanation, locking people out of years of work.
Creators know the exposure. Seventy percent say a single algorithm change could seriously hurt their income, and 77% report real income risk tied to platforms they do not control. The dependency is not a feeling, it is a line item.
There is a data dimension too. Eighty-five percent of publishers say first-party data will be the most significant driver of ad revenue by 2026. You cannot collect first-party data on a channel you do not own. Renting reach and renting the relationship turn out to be the same problem.
When you build on a platform, you are not building an audience. You are renting access to one, on terms that can change without notice.
It would be easy to read the decline of search and social reach as a decline in attention itself. It is not. People are on their phones more than ever. The attention simply moved to a place most businesses have not followed it to: inside apps.
The split is stark. Around 90% of mobile time is spent inside apps, with browsers and search taking under 6% of smartphone time, per DataReportal and data.ai. In the United States, roughly 70% of all digital media time happens in apps. People spent 5.3 trillion hours in apps worldwide in 2025, with about $167 billion in in-app spending and non-gaming categories overtaking gaming.
Apps do not just hold attention, they convert and retain it better. App visitors convert at checkout roughly three times more often than mobile web visitors, and apps hold about 32% retention against the mobile web's 20%. People keep apps on their phones. They forget browser tabs.
An app solves the two problems search and social created. It gives you a place on the home screen, permission to reach back out, and data that belongs to you. The most powerful of the three is the second one. A direct line to reach people again, on their terms, is what changes the economics.
The notification data is striking. Airship, studying 63 million users across 1,500 apps, found roughly 3x higher retention for users who receive at least one notification in their first 90 days. Users getting weekly notifications retained 440% better than those getting none, and daily notifications pushed that to 820%. Sixty-five percent of users return when push is on, and opt-in users show roughly 2x the retention and 4x the engagement of those who never enable it.
People are more receptive than the cynicism around notifications suggests. Sixty-eight percent say they are comfortable with daily or more frequent news and information notifications. Push open rates run about 50% higher than email, and a single notification in the first week can lift retention by more than 70%.
More is not better. Over-sending backfires: people start disabling notifications, or uninstalling, once they pass roughly six a day. Relevance and cadence are the whole game. Getting them right is a design problem, not a volume problem, and it is the part most businesses get wrong.
News and information is one of the most welcome notification categories there is. That is what makes a content app one of the best habits you can build.
An owned channel is not just an app icon. It is a system. Across the products we have built and studied, five parts show up every time the move works.
The objection used to be cost. Building a production-grade app meant a long timeline and a budget out of reach for a creator or a small business. AI-native development has changed that math. A real, shippable app can now be built in weeks, at a fraction of the old budget. The owned channel is no longer reserved for companies with an engineering team.
We did not arrive at this thesis from the sidelines. We ran it on ourselves, with a product whose entire reason for existing was being threatened by the shift this brief describes.
ThisWeek.AI began as a newsletter in 2022, with an audience built the slow, honest way. As AI search started absorbing the search traffic publishers depend on, we made the move we now make for clients. We built a native iOS app with AI-native development, designed it around a daily news habit, and gave readers a reason to return that does not pass through a search engine.
The result is an owned channel. Readers get their news from the app and come back through notifications, on their own schedule, on a surface we control. The relationship is first-party, the data is ours, and the audience does not depend on a Google ranking that could change next quarter.
Hamtech is an AI-native solutions studio. We build the apps and the cross-platform presence that give a business a direct line to its audience, and we build the product and the reach together rather than bolting distribution on after launch.
We work with three kinds of clients, and we start with the first. Publication and content creators, where the pain of disappearing search traffic is sharpest. Entrepreneurs and founders, who need a product and an audience at the same time. And small to mid-sized businesses, who want to stay top of mind on the surface their customers actually use.
In every case the job is the same: solve the technology problem and the distribution problem in one move, so the thing you build comes with a way to reach the people it is for.
Figures in this brief come from independent research published through 2025 and early 2026. Sources are listed below.