AI is disrupting all manner of industries, and one of the oldest technologies on the web — Google search — doesn’t seem to be immune.
Y-Combinator CEO Garry Tan has said that he’s heard from companies that referral traffic from Google is down 15 percent this year, likely because of changed user behaviour due to the proliferation of AI. “I don’t think it’s out there in the annual reports yet. And we did some research prior to this episode, we couldn’t really find anything that conclusive. But maybe purely anecdotally, because we’re in this pool of people who are very, very early adopters, very much software engineers, our behavior interacting with the internet has changed already,” he said on a podcast.
“It’s not a surprise to me. Some people are starting to report in their referral traffic — Google referrals are down maybe 15 percent in the last year. And that certainly probably mirrors my own behavior. Like, I still use Google but I’m increasingly not clicking on any links in Google because they’re sort of the snippet at the top or the first thing I think of is using ChatGPT with web or using Perplexity directly,” he added.
Millions of websites on the internet rely on Google to send them traffic. People search on Google, and are then directed to websites which might have information relevant to what they were looking for. But AI has upended this for many websites — a lot of information is increasingly being shown directly on Google’s search page as an AI-generated snippet, which means that people no longer have to click on the links of external websites, and can get their answers from the AI-generated snippet itself.
This isn’t the first time that Google has used AI in its search results. For a while, Google has been answering some search queries with direct result on its search page. But this has become much more pervasive — and effective — with LLMs. These days Google gives long and comprehensive answers to questions the user has asked, including itineraries, recipes, and other general information. Thus far, much of these questions brought traffic to company websites, but with Google answering questions right there on the search page, websites seem to be seeing reduced traffic. Also, there are other AI tools that people are using for things they previously used search for, like ChatGPT and Perplexity, which could be further eating into Google-generated traffic.
And this could have interesting consequences not only for websites but also for AI models. If websites stop getting traffic from Google, this could cause many websites to shut down, and could disincentivize newer websites into entering the content creation game. But AI models are ultimately trained on articles written by these websites, and AI companies have already used most of the existing data to train their models. And if websites don’t create new human-generated content, it could hamper the research efforts of AI companies to create better AI models. It’s a chicken-and-egg problem, and one that companies like Google and OpenAI will have to solve as they look to not only making their users happy with better results to their questions, but also ensure that internet websites keep creating human-generated content to train their AI models.