Lesson 5 of 6 · 8 min read · beginner
AI search and the future
AI answers are changing how people search and click. See what the data shows, what it means for feeds built on clicks, and where ads in AI assistants stand.
Search arbitrage rests on a habit: a person searches, sees a list, and clicks. For twenty-five years that habit was stable. Since 2024, search engines have been placing AI-written answers above the list, and standalone AI assistants now answer questions without showing a results page at all. This lesson looks at what is measured so far and what remains uncertain.
What AI answers change
For twenty years a search produced a page of links and ads (a SERP), and the searcher clicked through to a website. Search advertising, and search arbitrage with it, is built on that click.
- The old path: query, list, click: For twenty years a search produced a page of links and ads (a SERP), and the searcher clicked through to a website. Search advertising, and search arbitrage with it, is built on that click.
- An answer appears first: With AI Overviews and other generative search, many queries now return a written answer above the links. The searcher can read it and stop.
- Fewer clicks leave the page: A Pew Research Center study of US browsing in March 2025 found people clicked a result link on 8% of Google visits that showed an AI summary, against 15% without one. Google disputes the method. More searches end as a zero-click search.
- Ads move into the answer: The auction has not gone away. Google has placed ads in and around AI Overviews and in its AI Mode, so the same advertisers can appear inside the answer instead of beside the links.
- One budget, more places to spend it: An advertiser’s budget is finite. Each dollar spent inside an engine’s own AI answer is a dollar that did not need a partner site. Syndication exists because engines wanted extra reach; if they need less of it, feeds shrink.
- What it means for arbitrage: Arbitrage visitors mostly come from paid social and native ads, not from Google results, so the first click is less exposed than it is for sites living on organic traffic. The exposure is on the selling side: the feed is one supplier’s decision. That is concentration risk.
- Ads inside AI assistants: Chat assistants have begun carrying ads too; OpenAI started testing ads in ChatGPT in the US in early 2026 (ads in AI assistants). Whether any of them will syndicate ads to outside publishers, as search engines did, is an open question.
- Plan for a shorter chain: Nobody knows how far this goes. The safe reading for an operator: treat today’s feed economics as rented, keep costs flexible, and build pages people would value even with no ad on them. Platform risk is the business model’s main risk.
What has changed on the results page
AI Overviews are the summaries Google shows at the top of many results pages. Google said in July 2025 that AI Overviews had more than two billion monthly users. It also offers AI Mode, a conversational version of search. Microsoft has Copilot built into Bing. These are forms of generative search: the engine writes an answer instead of only listing links.
The effect on clicking has been measured. The Pew Research Center tracked the browsing of 900 US adults in March 2025. When a Google results page included an AI summary, users clicked a traditional result link on 8% of visits. When there was no AI summary, they clicked on 15%. Users clicked a link inside the summary itself on about 1% of visits, and were more likely to end their session after a page with a summary: 26% against 16%.
A search that ends without any click is a zero-click search. These existed before AI, for weather, sums and quick facts. AI summaries extend them to questions that used to need a visit to a website.
What this does, and does not, change for arbitrage
It helps to separate the two sides of the arbitrage trade.
Less exposed
- Operators who buy traffic on social or native platforms do not depend on Google's own results page for visitors
- The search feed on their page shows sponsored results, and those still exist
- Commercial intent queries, where people want to compare and buy, are where ads remain most useful
- Advertiser demand for measurable clicks has not disappeared
More exposed
- Anything relying on organic traffic from search engines, such as content sites
- Search-to-search models that buy search ads to send people to another results page
- Metasearch engine and alternative search sites whose users may move to AI assistants
- Informational verticals, where an AI answer satisfies the question completely
The deeper risk is indirect. Syndicated feeds exist because search engines chose to extend their ads to partner sites. If more searching happens inside AI answers that the search engines fully control, they have less reason to share advertisers with outside pages, and more reason to keep quality high where ads do appear. The policy tightening of 2024 to 2026 was not caused by AI, but it points the same way: fewer, stricter partners.
Ads inside AI answers
Advertising is following users into AI products. Ads in AI assistants is the general term.
- Google shows ads within and alongside AI Overviews and has been extending ads into AI Mode. These are bought through ordinary Google Ads campaigns, so advertisers do not need a separate product.
- Microsoft places ads in Copilot conversations through Microsoft Advertising.
- OpenAI announced in January 2026 that it would begin testing ads in ChatGPT in the United States for users on its free and Go tiers, with paid tiers above that remaining ad-free. Press reports say it stated that ads would be labelled and would not influence answers.
What does not yet exist at scale is the arbitrage equivalent: a widely available programme that lets a small third-party publisher embed AI-assistant ads on its own pages and take a share, the way a search feed does. Whether the AI companies will build open syndication, keep ads to their own surfaces, or work with a handful of approved partners is undecided. Anyone who tells you how this ends is speculating.
AI on the operator's side
Operators use AI too. AI-generated content makes articles and ad variations cheap to produce. That cuts costs, and it also lowers the barrier for everyone, so the advantage is brief. It carries risk as well: Google's rules for Related Search on Content require pages to have value of their own, and mass-produced pages that add nothing are the kind that get labelled Made-for-arbitrage. AI is a production tool. It does not make a thin page acceptable.
How the listed companies are responding
Public companies give a view of where management teams are placing their bets. System1 said in 2026 that it had significantly reduced marketing tied to search monetisation and launched products aimed at AI agents. Perion has redirected its business toward connected TV, digital out-of-home and retail media. Team Internet has been reviewing options for its domains business while shrinking and reshaping its search segment. The common thread is reducing dependence on a single search feed.
Three plausible paths
| Path | What it would look like | Who it favours |
|---|---|---|
| Contraction | Fewer feeds, stricter terms, lower volumes, more operators leaving | Large, compliant partners with capital |
| Substitution | AI platforms open their ads to approved third-party sites | Whoever holds the first partner contracts |
| Specialisation | Arbitrage survives mainly in commercial verticals where people still compare options | Operators with strong content and clean traffic in those verticals |
Key takeaways
- AI Overviews and AI assistants answer questions directly, so fewer searches end in a click.
- Pew found users clicked a result link on 8% of visits with an AI summary against 15% without one.
- Operators buying social or native traffic are less directly exposed than models depending on search-engine traffic.
- Ads are appearing in AI answers at Google, Microsoft and OpenAI, but open syndication to third-party sites is not established.
- Listed arbitrage companies are reducing their reliance on search feeds; the long-term shape is uncertain.
Questions people ask
Will AI kill search arbitrage?
Nobody knows. AI answers reduce clicks on results pages, and search engines have been tightening partner programmes for separate reasons. Operators who buy social or native traffic still have sponsored results to show, especially for shopping-type queries. The likely near-term outcome is a smaller, stricter industry. Whether AI platforms open ad syndication to outside publishers is undecided.
Do AI Overviews reduce clicks?
The best public evidence says yes. Pew Research Center found that in March 2025, US users clicked a traditional search result on 8% of Google visits that showed an AI summary, compared with 15% of visits without one. They clicked a link inside the summary about 1% of the time and ended their session more often.
Are there ads in ChatGPT?
OpenAI announced in January 2026 that it would start testing ads in ChatGPT for users in the United States on its free and Go tiers, while higher paid tiers stay ad-free. According to press reports it said ads would be clearly labelled and would not affect answers. There is no established programme for outside publishers to carry those ads.
What is a zero-click search?
A zero-click search is one where the person gets what they need on the results page and leaves without clicking any result. Weather, calculations and quick facts have long worked this way. AI-generated summaries extend it to more complex questions, which reduces visits to websites and, over time, the number of opportunities to show a paid result.