Perplexity, Google AI Overviews, and ChatGPT Search are reshaping how we find information. Traditional blue links may never recover.
The search paradigm that has governed the internet for a quarter century, type a query, scan ten blue links, click through, is being replaced in real time by AI systems that simply answer the question, and the shift is now large enough to show up in publisher traffic dashboards, advertiser spend allocations, and the basic economics of who gets paid for producing information online.
Perplexity's growth curve is the clearest signal
Perplexity has become the category's clearest proof point, growing from 10 million to 100 million monthly active users in just 18 months, a growth rate that traditional search products have not matched at any point in the last decade. Its AI Search, built on Sonar and related model variants, synthesizes information from multiple sources into a single answer, attaches citations back to the original material, and handles natural follow-up questions in the same way a conversation would, rather than forcing the user to reformulate a fresh query each time. Users repeatedly describe the experience as having a research assistant who has already read the entire internet, which is a meaningfully different mental model from typing keywords into a search box and hoping the ranking algorithm surfaces the right page.
Google's AI Overviews are eating click-through rates
Google's answer, AI Overviews, now appears on more than 40 percent of searches, placing an AI-generated summary directly above the traditional list of results. Because that summary frequently contains enough information to satisfy the user's question outright, many searches never produce a click to any underlying source at all. Click-through rates to publisher sites have fallen 25 percent since AI Overviews rolled out broadly, a decline steep enough that it is now a standing agenda item at digital publishing conferences and a direct line item in falling ad revenue for sites that built their business model around organic search traffic.
The information economy is being renegotiated in real time
The structural shift underneath these numbers is that the entity answering the question, not the entity that originally published the information, is now the one capturing the user's attention and, increasingly, their trust. Publishers that spent two decades optimizing for search engine ranking are now exploring direct licensing deals with AI companies, trading a share of their content for a fee that does not depend on click volume at all. A new measurement category, AI citation rate, tracking how often a publisher's content is cited as a source inside an AI-generated answer, is emerging as a metric publishers now track with the same seriousness they once reserved for search rankings.
Why the follow-up question changes user behavior
Part of what makes AI search stickier than traditional search is the ability to refine a question in conversation rather than starting over. A user who gets an incomplete answer can simply ask a clarifying follow-up, and the system retains context from the original query, collapsing what used to be three or four separate search sessions into a single continuous exchange. That compounding convenience is a large part of why usage curves for AI search products have grown so much faster than any comparable search product in the traditional era.
What this means for anyone who relies on search
For researchers, journalists, and analysts who depend on being able to trust and verify what a search tool tells them, the citation mechanism is not a nicety, it is the entire basis for whether the answer can be used professionally at all. Tools that provide a direct answer without a clear, checkable source trail are far less useful for any work where getting it wrong has consequences.
Vincony's AI Search, powered by Perplexity's Sonar technology, extends this idea further by letting users search across more than 800 AI models simultaneously, compare how different models interpret the same query, and receive cited answers that link back to their original sources, making it one of the more comprehensive AI search interfaces available for anyone whose work depends on verifiable, sourced information rather than a single model's unchecked output.