AI is like an earthquake for digital marketing — every leader feels the ground shaking. The question isn’t whether change is coming, but how to respond: where to leap, how to stay upright, and ideally, how to emerge stronger. Science tells us that humans are descended from agile creatures who were able to seek shelter from the cataclysm that wiped out the dinosaurs. While AI may not be as apocalyptic as an asteroid blotting out the sun, the goal remains the same: ensure our companies are among the survivors.
The earthquake I’m referring to is the shakeup in search engine optimization (SEO), the industry where I’ve spent most of my career, and one that’s been subject to periodic quakes that happen every time Google changes its algorithm. But those were tremors compared with the arrival of ChatGPT and the other large language model (LLM) chatbots.
We’ve been watching the growth of traffic to ChatGPT since its launch in November 2022, and in 2024 it was among the world’s fastest growing websites in multiple markets around the world, as reflected in the annual Similarweb Digital 100 report. We’re seeing continued double-digit growth in its traffic, month after month, in 2025.
When consumers are looking for information on the web, they now choose between entering keywords into Google or prompting ChatGPT for an answer. Meanwhile, Google isn’t sitting still – it’s promoting Gemini as a direct competitor to ChatGPT and progressively introducing LLM technologies into Google search. Other search engines, including AI-centric ones like Perplexity, and LLM players like DeepSeek and Mistral, are seeking opportunities amidst the shakeup.
While the technologists battle, digital marketers are dismayed to find that proven strategies aren’t working like they used to. Meanwhile, they’re getting tons of advice about what to do next, much of it contradictory.
What does the data say?
While I don’t have all the answers, I am in the business of helping Similarweb customers discover where they stand in the digital marketplace and how to improve their position. We’re moving fast to keep pace with changes, and we now have powerful tools to help. We can measure how many referrals companies get from AI chatbots as opposed to search engines, how many Google searches trigger AI Overviews, and how consumer behavior is different for AI-mediated queries as opposed to traditional searches.
AI may even help solve the problems caused by AI – we’re currently working on AI agents that can do things like sort through all the varied data available from across our platform and recommend strategies for capitalizing on trends and avoiding pitfalls.
Meanwhile, the news isn’t all good, but it isn’t all bad, either.
Early results suggest that ecommerce players stand to benefit from AI-driven referrals to their websites – if they can master the trick of getting the AI bots to recommend them.
The conversion rate, or ratio of visits to sales transactions, for AI referrals is rapidly catching up with that for search, and was actually incrementally higher (6.38% vs 6.29%) around the December holidays this year. While this data reflects the UK market, showcased at April’s BrightonSEO conference, the trends in the US are very similar.
Over the last year, HomeDepot.com has quietly racked up 2.3M visits via AI chatbots — with ChatGPT driving a whopping 70% of it, followed by Perplexity.ai at ~29%. For the latest 6 months of data we analyzed, Home Depot has averaged 250K monthly referrals visits from ChatGPT.
At first, what we saw was that chatbots would send traffic to the front page of a recommended website. However, Home Depot is an example of where the traffic is landing on more specific product and category pages like refrigerators and kitchen appliances, outdoor irrigation gear, paint, tools, patio furniture, and lighting. These are high intent-to-buy visits, propelled by more detailed conversations with the AI chatbots.
Interestingly, users referred by chatbots appear more engaged: they spend more time on site (11 minutes for ChatGPT versus 8 for Google) and view more pages of content (13.3 versus 11.6). This is all favorable shopper behavior that makes a compelling case for AI referrals as a valuable source of digital traffic.
Tough time for publishers
On the other hand, publishers have good reason to be nervous about potential readers who never click through to their website because a chatbot answered their question, leaving little reason to click on a link to the article the answer was based on. That’s assuming the link is displayed, which it’s not always.
As part of research performed for the Axios Media Trends Executive newsletter, Similarweb found that traditional search traffic to the top 500 media and news websites in the U.S. dropped by more than 15% from an average of 5.3 million page referrals per publisher in May 2024 to an average of 4.5 million in February 2025. During the same period, AI chatbot referrals to those same publishers rose by more than 2,100%, but the volume of AI referrals is still miniscule in comparison.
One big question is whether the occurrence of AI Overviews in Google search results drives up zero-click behavior – users searching but finding enough of an answer on the search engine results page (SERP) that they don’t click on any link. Zero-click searches have been rising for years, at least partly as the result of other summarization features Google began adding long before the advent of AI Overviews.
Although it seems intuitively obvious that AI Overviews would accelerate that trend, at first we found no clear correlation. However, one confounding factor is Google may display an AI Overview for a given keyword or phrase one day and not the next. When we controlled for keywords that triggered an AI Overview 80% of the time or more, the rise in zero-click behavior became more pronounced.
AI driving zero-click behavior higher isn’t just an issue for publishers but for every organization that has SEO and content marketing as a component of its marketing plan.
How to survive the earthquake – and thrive amid disruption
Keeping in mind that the rules of the game are continually being rewritten, we’ve published tips and strategies for AI search from which I can share a few highlights:
Discover and track AI referrals. Knowing which pages of your website attract chatbot referrals will tell you what content to produce and enrich to get more of that traffic. Knowing which popular prompts are associated with AI referrals may soon be as important as traditional keyword optimization.
Likewise, knowing which keywords that have traditionally been traffic drivers are now triggering AI overviews is the first step to responding to that trend.
Drive up your search ranking. While that’s easier said than done, it makes a difference in the context of AI Overviews. The websites that rank highest in traditional search are also the ones most likely to be cited (and linked to) as sources of information in the body of an AI Overview.
Increase brand visibility and credibility. Building a strong online presence through backlinks and brand mentions helps AI systems understand and rank your content effectively. AI engines are looking for authoritative sources of answers to user questions. Seek to be on that list.
Make your website AI-friendly. This includes established technical SEO techniques like adding sitemaps and structured content, but you can also take advantage of more specific tools like Wordlift’s LLM.txt generator.
Identify search and prompt intent. Even if AI results in fewer visitors to your website, if it helps attract the right visitors – like those high intent-to-buy visitors coming to an ecommerce site – it may turn out to be a net positive. Just make sure you provide clear, concise answers to the questions those users were asking in the first place.
Stay tuned
The earthquake is not over, and we don’t know whether the biggest shocks are yet to come. The best thing you can do is pay close attention and monitor the performance of your websites versus those of your competitors. Take advantage of the disruption and avoid the pitfalls as best you can, but don’t count on the pace of change slowing down.
