To preserve the usefulness of its rankings as early adopters defect to ChatGPT, Bing, and Perplexity, Google just completed rolling out the largest algorithm update in the company’s history. But search marketing experts and business owners say spam is still commanding top-ranking positions, calling into question Google’s ability to deliver the best search results in the brave new world of generative AI.
“I’ve never seen as much spam in Google’s results so blatantly in top positions,” says Lily Ray, VP of search strategy and research at digital marketing agency Amsive. “I’ve never seen so many big companies getting away with publishing really low-quality content that’s ranking prominently. Probably 20 to 30 people every day message me desperately because either their websites are being outranked by spam or AI content, or their content is being scraped and plagiarized by AI tools.”
Google derives more than 75% of its revenue from selling ads on search engine results pages (SERPs). That revenue is reliant on the company’s ability to deliver trustworthy, useful search results. As long as it continues to deliver the goods, it has an audience. But if search result accuracy deteriorates, audiences may defect, and advertisers could follow. So defending the reliability of search results against AI abuse is critical to Google’s future prospects.
Google vs. search spam
When I asked Google for a statement about how artificial intelligence is impacting its search business, a spokesperson provided the following: “Our research shows that [Google] Search satisfies the overwhelming majority of user needs for people around the world. We’re continuously improving our advanced spam-fighting systems to keep Search more than 99% spam free and to address new types of low quality content, including AI-generated spam. Our major updates in March were an example of this.”
Google finished rolling out these algorithm updates on April 19, so their impact may only now be fully realized. But Google’s traffic dipped more than 5% from 172 billion visitors in January to 163 billion in February, bouncing back to 165 billion in March, according to estimates from Semrush. And an unverified, damning report by PR consultant and writer Ed Zitron alleged that management pressured engineers to find ways to prolong session times and boost search activity, even if it meant degrading the user experience.
It’s important to recognize that marketers who critique Google from afar may not be impartial, as they could be motivated by the prospect of attracting new clients through increased visibility. But it should also be noted that Google management “discussed increasing search queries on Chrome to improve ad revenue,” according to Search Engine Land.
During its Q4 2023 earnings report, Google’s parent company Alphabet did not disclose total search volume figures. The company focused on broader financial and operational performance, product enhancements, and strategic initiatives, particularly around AI integration into various services, including search.
According to a CNBC article, Google search chief Prabhakar Raghavan told employees that users “may have a new gizmo out there that people like to play with, but they still come to Google to verify what they see there because it is the trusted source, and it becomes more critical in this era of generative AI.” Semrush Trends estimates that roughly 40% of the traffic from ChatGPT and 20% of the traffic from TikTok goes straight to Google.
We’ve all grown more skeptical of algorithms. Before AI chatbots enabled spammers to autogenerate topically relevant web pages at scale, Google’s algorithmic results were more widely regarded as trustworthy, impartial, and tamper-proof. But according to chatter in the SEO forum on Reddit, that opinion is in flux as a new crop of web pages created with generative AI tools ascend to the top of organic rankings in the SERPs as well as Google’s latest experiment, the Search Generative Experience.
If you’re unfamiliar with SGE, it’s like featured snippets—the brief pieces of information that sometimes appear at the top of the results page—on steroids. The feature is still in beta and available only to a small percentage of users. It’s Googe’s attempt to leapfrog ChatGPT by coupling search with its Gemini large language model to generate similar results in a more graphically elegant, scannable way, but with links to the sources it lifts information from. ChatGPT offers no attribution.
“SGE results vary depending on what type of query it is,” says Mike King, CEO at SEO agency iPullRank. “If it’s a shopping query, Google presents a bigger experience that takes up all the screen space above the fold and gives you things like comparisons on products. The local experience is also different in that they’re trying to help you make a purchasing decision—like where to go for pizza—right there in the search results.” Both scenarios will reduce the traffic Google sends to websites.
Optimizing for AI
According to researchers at Princeton and Georgia Tech, AI “is rapidly replacing traditional search engines like Google and Bing.” Just as websites can be optimized for search engines, they can also be optimized for LLMs through an emerging process called generative engine optimization. The researchers’ study, which includes sample prompts, found that the use of quotations, statistics, and writing in an authoritative tone made content more likely to be cited in responses generated by large language models like ChatGPT.
It has never been easier to learn how to trick the Google algorithm. You can game Google search with AI by watching YouTube videos. Run a simple query for “seo with ai,” and you’ll see dozens of tutorials on how to pollute the internet with web spam. Ironically, some videos are about how to use Google’s Gemini to manipulate Google’s search rankings.
Generative AI may not be the only reason so many people are complaining about Google’s March core update. “Too many people understand how Google works,” says Tim Soulo, CMO at Ahrefs. But SEO specialist Clark Mackey disagrees. “The tools have improved, but there are so many potential failure points that SEO has actually become much more challenging and nuanced,” he says.
But AI makes it easier to overwhelm the Google algorithm by mass-publishing topically relevant, keyword-stuffed noise. The abusers spin up expansive networks of these spammy web pages and interlink them to try and spoof the patterns Google monitors to determine rankings.
To address the generative AI abuse problem, Google is favoring content from sources that are more likely to be human-generated, lifting Reddit and Quora discussions to the top of the SERPs. That means user-generated content from these sources about topics such as financial and medical information are more likely to appear on page one.
In a world increasingly dominated by generative AI text, forums are a treasure trove of authentic, human-authored content. In fact, Google recently agreed to pay $60 million to Reddit to train its large language model on Reddit’s reviews and testimonials. But forum comments also have a tendency to be insidious and polemic in nature. Their reliability varies widely, and it’s harder to tell whether authors have conflicting interests that might bias their opinions because you can’t see them like you can on Instagram and TikTok.
“Websites like Reddit might not be the best training set for AI, if you define quality as facts supported by evidence,” says Ted Kubaitis at SEO Tool Lab.
“Google is shifting its responsibility for maintaining the quality of results to moderators on Reddit, which is dangerous,” says Ray of Amsive. Search for “kidney stone pain” and you’ll see Quora and Reddit ranking in the top three positions alongside sites like the Mayo Clinic and the National Kidney Foundation. Quora and Reddit use community moderators to manually remove link spam. But with Reddit’s traffic growing exponentially, is a human line of defense sustainable against a generative AI bot army?
Google fails the Turing test
It appears as though Google is unable to reliably tell the difference between human-generated and generative AI content. In all fairness, it’s not alone—so marketers are having a field day exploiting AI to try and commandeer rankings. By clogging up the internet with keyword-stuffed crap, they’re making it tougher and more expensive for Google to separate the signal from the noise. Crawling the web is expensive, so crawling crap is flushing money down the drain.
SEO has become a critical piece of the digital marketing puzzle. Google has guidelines about what is and is not prohibited. But they’re about as useful as the Fed chairman’s remarks on interest rates—carefully wordsmithed and decidedly vague. They have to be. Web content is not uniform, Google’s systems are susceptible to manipulation, and the company has to be cognizant not to throw the baby out with the bathwater by making algorithm tweaks that make it tougher for high-quality websites to be found in search results.
Consequently, Google publishes loose guidelines it can use to manually deindex offenders—according to website owners who have seen as much as 90% of their traffic evaporate in some cases—and there’s no legal recourse, due process, transparency, or oversight. If Google makes an algorithm tweak that puts a company out of business, so be it. The prevailing consensus on X (formerly Twitter) is that small businesses are getting hit harder by the latest algorithm update than larger ones.
That’s the risk of relying too heavily on organic search traffic. Google prioritizes the needs of searchers over businesses, emphasizing the importance of diversifying traffic with sources such as social media and chatbots.
Search fragmentation means diversification
There’s a generational shift happening in the way people search for information, and it’s fragmenting the search experience. “What happens when a brand is no longer a verb?” King asks. “My kid’s babysitter doesn’t just Google. She searches it up.” In other words, she consults multiple sources.
Google is still the main event for search, with 165 billion visits in March. But it’s no longer the only game in town. For perspective, during the same period ChatGPT logged just 2.5 billion visits, TikTok had 3.6 billion, Instagram counted 8 billion, and Reddit had 7.8 billion. That’s just on the web and doesn’t include mobile app usage, which is much harder to track.
This fragmentation of the search experience, which gives individuals and organizations alternatives to Google, is a positive development, because it introduces more competition, which drives innovation and economic growth. But generative AI could be a Faustian bargain for Google search.
Google must embrace AI or risk obsolescence. But it also must defend its search rankings from generative AI abuse. So the company is adding generative AI capabilities to services like Gmail and Google Docs to reduce our need to write manually while attempting to filter out that very same content from its search results.
The company is telling us something by hosting YouTube videos on how to hack its search results. Either it’s unable to detect and remove these videos at scale, or it’s too reliant on the ad revenue they generate. Much as the migration to the internet forced the media businesses to trade their analog dollars for digital pennies, it may now be Google’s turn to burn the boats.