Google’s first year of AI-powered search was an exciting ride. It will only get wilder.
March 2024 was a unique update. It was one Google’s biggest core updates, and it took more than a month for the update to be fully rolled out. Since then, nothing has felt the same. It depends on who asks and who you are.
Websites often see traffic changes following a core update. But the impact of the update in March 2024 was a seismic shift. Google claims that the update was designed to tackle spam and AI generated content in a meaningful manner. Despite this, many publishers claim that they have seen clicks on their legitimate sites disappear, while others are dealing with unprecedented volatility in traffic. Google controls almost all of the search market. Changes in its algorithm could have a major impact on the Internet.
Looking back, the March 2024 algorithm update appears to be the first major Google update in the AI era. It (supposedly) veered away from ranking AI-authored online content, but it also laid down the foundation for Google’s ambitious – and often annoying – desire to fuse AI with its search.
This ambition was first revealed with AI Overviews a year ago. Now, the company is taking a more audacious approach, adding a chat-based answer system called “AI Mode.” These technologies aim to do two things at least: They want you to stay on Google properties for longer and they remix publisher’s content without always giving prominent references.
It appears that smaller publishers have been the ones to bear the brunt of these updates. “Google got all this flak for crushing the small publishers, and it’s true that when they make these changes, they do crush a lot of publishers,” states Jim Yuis the CEO of enterprise SEO platform BrightEdge. Yu explains that Google will be the only search engine to surface niche content, and sites on the fringes are bound to change during a major core upgrade. Google’s own assessment of the impact of March 2024 is surprisingly positive. The company stated that it hoped to reduce the appearance unhelpful content on its search engine result pages (SERPs), by 40 percent. The company claimed that after the update, it had actually reduced the amount of unhelpful content by 40%. Closer to 45 percentDoes it feel like Google has improved that much? Most people don’t believe it. What is the cause of this disconnect? According to Michael King– founder of SEO company iPullRank – says that we don’t speak the same language as Google. He says “Google’s internal success metrics differ from user perceptions,” . “Google measures user satisfaction through quantifiable metrics, while external observers rely on subjective experiences.”
Google tests algorithm changes using various tests including human search quality testers, and running A/B testing on live searches. But success is more important than anything else. Total number of searches (five trillions per year). This number is often the centerpiece of Google’s business updates, to show investors that they can still grow.
Using search volumeas a measure of [quality[quality is problematic. If a search engine is getting more attention, it could mean that the quality has decreased so people will try different queries (e.g. the old trick to add “Reddit” at the end of the search string). People could be searching because they are unhappy with the results. Jim Yu says that Google may be moving too fast and breaking things but it’s not as bad as people think. Yu. “[Google] is held to a higher standard, but by and large, I think their search quality is improving.”
King says that Google’s search behavior favors the big names but that other sites are starting to see a recovery. “Larger brands are performing better in the top three positions, while lesser-known websites have gained ground in positions 4 through 10,” says King. “Although some websites have indeed lost traffic due to reduced organic visibility, the bigger issue seems tied to increased usage of AI Overviews”-and now the launch AI Mode.
The specter AI is hanging over every SERP. The unhelpful feeling that many people get from Google search, regardless of internal metrics the company uses, may be due to a fundamental change in how Google surfaces information at the age of AI.
AI Overview hangover
By 2025, it’s impossible to talk about Google’s search changes without mentioning the AI-generated elephant. Google announced that it would expand AI in search as part of its core update in March of 2024. The “Search Generative Experience” was moved from the labs to Google.com. The AI Overview box, formerly known as “AI Overviews.”
has been a staple on Google’s results page since its debut one year ago. The feature uses the exact same AI model as Google’s Gemini chatbot in order to formulate answers to search queries. It does this by ingesting 100 (!) of the top search results. It sits at the top of the page, pushing so-called blue link content even further down below the ads and knowledge graph content. It sits on top of the page pushing the so-called Blue Link below the ads and Knowledge Graph content. It doesn’t respond to every query and sometimes answers questions that you didn’t even ask, or gives a completely wrong answer.
It’s not without irony that Google’s laudable move to de-rank artificial AI slop occurs at the same time as Google heavily promotes AI-generated content at the top SERPs.
AI Overviews are displayed at the top of most search results. Google
How much does Google get for all this AI work? It would appear that more eyeballs are being attracted. “AI is driving more engagement than ever before on Google,” Yu says. BrightEdge data indicates that Google impressions have increased by nearly 50% since AI Overviews was launched. AI Overviews is often criticized online, but this doesn’t mean that people don’t care about the feature. Google reported in its Q1 2025 earnings that AI Overviews was “used” every month by 1.5 billion users. This “usage” statement should be taken with caution, as you cannot easily opt in or out of AI Overviews.
It’s interesting to note that the impact of AI Overviews varies across the web. Google was so happy with AI Overviews in October 2024 that it expanded their use to more queries. As AI began to appear in more queries, publishers experienced a corresponding drop in traffic. Yu estimates that this drop is around 30 percent in average for those who have high AI query coverage. The traffic change is negligible for searches that are not supported by AI Overviews, such as financial services and restaurants. There are always exceptions. Yu suggests that large businesses with high AI Overview coverage have seen smaller drops in traffic, because they rank highly as both AI citations AND organic results.
Lowering traffic isn’t the death of some businesses. AI Overviews were absent from B2B searches in May last year, but this has changed dramatically in recent months. BrightEdge estimates 70 percent of B2B queries now have AI responses, which has led to a reduction in traffic for many businesses. Yu does not think that it is all bad. “People don’t click through as much—they engage a lot more on the AI—but when they do click, the conversion rate for the business goes up,” Yu said. Theoretically, serious buyers will click while window shoppers won’t.
The Internet is not just a giant shopping mall for shoppers. It is a place where people can share and find information. AI Overviews has hit some information providers hard. AI Overviews was heavily focused at launch on “What is” or “How to” searches. Bloggers and large media use “service content” to drive traffic. They don’t care about sales conversions. AI Overviews “helpfully” remixes and repackages their content, removing the need to click to the site. Some publishers are indignantly angry, asking why Google should remix content that it doesn’t control, without compensation.
Google’s intentions do not end with AI Overviews. Last week, Google began a public test of “AI Mode,” from the front page. AI Mode does not bother with the blue links. It’s a chatbot that, for now, attempts to answer your question without clearly citing the sources inline. It may mention Reddit, Wikipedia or other sources. Google offers a small box on the right-hand side of the screen with three links. You can expand the box to see more options. It’s not clear to the end user if these are “sources,” “recommendations,” and “partner deals.”
. Perhaps even more surprising, in our testing we found that not one AI Mode “sites box” included a site which ranked on the top page for the same query when performing a regular Google search. The links in AI Mode “best foods to eat for a cold” do not overlap with the SERPs for the same query on Google Search. AI Mode is a new feature, so its behavior will likely change. The company’s direction is clear.
Google’s real goal is keeping you on Google and other Alphabet products. Rand Fishkin noted that Google’s 2019 strategy is to keep you on Google or other Alphabet properties. The evolution from search engine into walled garden had reached a tipping-point. For the first time, more than half of Google’s searches led to zero click-throughs. The data showed that Google’s properties like YouTube and Maps were the most popular. You wouldn’t know if Google didn’t intend “zero-click” to deliver a search experience based on historical performance data or new features developed by the company.
The AI Overviews also don’t reveal this. They do cite the sources that were used to create each output and data suggests that people click on th ese links. But are the citations correct? Has every source that was used to create an AI Overview been cited? We don’t know because Google is notoriously opaque when it comes to how its search engine works. We do know, however, that Google uses a customized Gemini version to support AI Overviews. Gemini has also been trained using billions and millions of webpages.
It’s not always clear how AI Overviews cites a source. AI Overview’s output, as evidenced by the many hallucinations that we’re all familiar with (telling people they should eat rocks for example), is not good. We only know that Google is not transparent about this.
There are no signs of slowing down
Google is still pushing AI in search. Recent core updates have solidified this new arrangement, with an increasing number of AI-answered questions. The company seems to be comfortable with its current accuracy issues, or at least it is confident enough to push out AI upgrades anyway. Google seems to have been caught off guard by the launch of ChatGPT and is now using its search dominance in order to catch up.
To complicate matters further, Google doesn’t evenattempt to address the most important issue in this whole situation: the company’s quest for zero click search harms content creators on which it has built its empire.
Google, on the other hand, has been celebrating AI developments and insisting that content creators don’t know what is best for them. They have also refuted any concerns by making comments about increasing search volume and more complex search query strings.It’s obvious that the changes are working!
Google is building towards this moment since years. The company began with 10 blue links, and nothing else. But, little by little, they pushed those links down the page, and added more content to keep people in the Google ecosystem. Google introduced Universal Search in 2007. This allowed the company to insert content from Google Maps and YouTube, among other services. Rich Snippets started displaying more information from search results in SERPs in 2009. In 2012, Knowledge Graph began displaying answers in search results by extracting data from the search results. Each change increased the time people spent on Google and reduced click-throughs while also pushing the search results to the bottom of the page.
AI Overviews and AI Mode are the logical result of Google’s long-term transformation from an information indexer to a web portal that scrapes content from the web. In the early stages of Google’s development, websites were implicitly agreeing to allow Google to crawl pages in exchange for traffic. This relationship has strained over the years as Google has retained more traffic and reduced click-throughs. Locking out Google is not a realistic solution, since the company controls nearly the entire search market.
Business concerns can interfere with Google’s friendlier approach. During the antitrust trial for search, Documents showed that Google originally intended to allow sites to opt out of being trained for AI for its search-based AI feature–but they would still be included as search results. The company eventually canceled that idea, leaving website operators with a Pyrrhic option of either participating in AI “revolution” and becoming invisible on web or becoming invisible. Google is now competing with the open web, not supporting it.
Many of us feel a strange vibe when we look at Google’s results today. It could be the AI, Google’s algorithm or the Internet isn’t as it used to be. The shift to zero-click search, which began over a decade ago, was made evident by the March 2024 update. It has only intensified with the launch AI Mode. Even businesses that did not experience major traffic drops due to AI Overviews may soon find that Google’s AI only search can be much more intrusive.
AI slop continues until morale improves.
Ryan Whitwam is a senior technology reporter at Ars Technica, covering the ways Google, AI, and mobile technology continue to change the world. Over his 20-year career, he’s written for Android Police, ExtremeTech, Wirecutter, NY Times, and more. He has reviewed more phones than most people will ever own. You can Follow him on Bluesky () where you can see photos of his mechanical keyboards.
