Anastasia Braitsik is a titan in the global SEO and content marketing landscape, having navigated the digital shifts of the last two decades with a focus on data-driven precision. As a leader in data analytics, she has built a reputation for seeing past the hype of “the next big thing” to identify the structural changes that actually move the needle for businesses. Today, as the search market fragments into a multi-layered ecosystem of large language models and integrated AI search experiences, her expertise is more critical than ever. In this conversation, we explore the redistribution of search traffic, the strategic nuances of enterprise-focused platforms like Claude, and why the current obsession with a single leaderboard might be leading marketers astray.
The interview covers the recent volatility in AI referral shares, specifically focusing on the decline of Perplexity compared to the rapid ascent of Gemini. We discuss the historical parallels between the current AI landscape and the early 2000s search engine wars, the emergence of a three-tier tracking system for SEO professionals, and the necessity of distinguishing between consumer-facing chatbots and AI layers embedded within dominant ecosystems like Google.
Perplexity’s referral share recently dipped below 5% while Gemini surged past 10%. With some experts suggesting we should drop Perplexity from our trackers entirely, how do you view the strategic importance of smaller players in such a volatile market?
The numbers are certainly striking, and I can see why there is a push to simplify our reporting. When you look at the StatCounter data from this year, Perplexity’s referral share dropped from a healthy 7.91% in June down to just 4.31% by August, while Gemini climbed to 10.9% in that same window. It creates this feeling of a shrinking platform, but I believe we have to be very careful about declaring any player “irrelevant” based on a few months of volatility. If you look at the broader web visit data from Similarweb, Perplexity and Copilot both sit around 1.3% of worldwide visits, which is tiny compared to ChatGPT’s 53.9%, but “small” and “irrelevant” are not the same thing in a strategic sense. If you remove Perplexity today, you risk blinding yourself to a platform that still holds a 90% citation rate for some specific brands, even if its total traffic volume is lower than the giants. We have to move away from the idea that every platform deserves equal weighting in our trackers, but deleting them entirely is a reactionary move that could cause us to miss the next major shift in how users discover information.
You’ve mentioned that this current debate feels like a flashback to the early 2000s. What did that era teach you about the risks of focusing only on the “major” players of the day?
I remember sitting in a conference in Boston back in March 2002, just as the industry was still licking its wounds from the dot-com crash. At that time, everyone was convinced that the search market was settled and that we should only care about the big five: Yahoo, Excite, Lycos, AltaVista, and Ask Jeeves. I distinctly remember pointing out that Google was missing from that list, even though they were gaining share at a blistering pace and offering a fundamentally better experience. The mistake people made then was assuming that today’s incumbents would be tomorrow’s leaders, and we are seeing that same psychological trap today. If we only track the top three or four models, we ignore the outsiders who are iterating faster or serving specific niches more effectively. The search landscape is never truly “settled,” and the moment you stop watching the platforms that are gaining ground—even if they are starting from a small base—is the moment you become obsolete as a strategist.
We are seeing a significant redistribution of attention, with ChatGPT’s dominance slipping from over 76% to around 52.7% over the past year. Does this suggest the market is finally moving toward a more balanced oligopoly?
It’s less of a winner-take-all scenario and more of an emerging three-tier market where different players leverage different distribution advantages. ChatGPT is still a powerhouse with over 1 billion active users across its products, but its share of web traffic has definitely been challenged as Gemini and Claude find their footing. Gemini has the massive advantage of being baked into the Google ecosystem, reaching 1 billion monthly users and leveraging the world’s most dominant search infrastructure. Meanwhile, Claude has carved out a very specific, high-value territory in enterprise and professional workflows, with Anthropic reporting that they’ve already surpassed $65 billion in annualized revenue. This isn’t just three companies fighting for the same piece of pie; they are building three different types of businesses. We are seeing a core group of ChatGPT, Gemini, and Claude solidify their positions, but they are doing so by appealing to different user needs and distribution channels.
Claude seems to be a special case because its consumer traffic is relatively small, yet its enterprise adoption is massive. How should B2B marketers weigh a platform that doesn’t necessarily send a ton of “public” referral traffic?
If you only look at consumer traffic, you are going to massively undervalue Claude, which would be a catastrophic mistake for any B2B or SaaS company. Anthropic has been incredibly aggressive with enterprise distribution, landing massive deals like training 30,000 professionals at PwC and making the platform available to 50,000 employees at TCS. When you have over 1,000 business customers spending more than $1 million annually, you are looking at a platform where high-level decisions are being made and research is being conducted, even if that activity doesn’t show up in a standard referral report. A cybersecurity vendor or an enterprise software firm needs to care way more about their visibility on Claude than a B2C retailer might. This is why a one-size-fits-all tracking portfolio is dead; you have to connect your visibility metrics to where your specific audience actually lives, whether that’s in a professional dev environment or a casual mobile app.
Google’s AI Overviews and AI Mode are often lumped in with LLMs, but you’ve argued they should be treated as a separate category. Why is that distinction so important for SEO professionals?
Treating Google’s AI features as just another LLM is like treating an ocean like a swimming pool—the scale is just fundamentally different. Google reported that AI Overviews are reaching more than 2.5 billion users per month, and AI Mode queries have been doubling every single quarter since they launched. By May, AI Overviews were appearing in 43% of all U.S. searches, which is a massive jump from the 15% we saw only a year ago. This isn’t a niche chatbot experience; it’s an AI layer that has been integrated into the world’s primary discovery engine. For an SEO, the goal isn’t just to measure an LLM’s response; it’s to understand how AI is fundamentally re-engineering the entire search journey from start to finish. If you average Google’s AI performance into a generic “LLM score,” you lose the signal of how the dominant search provider is changing the way 2.5 billion people find information every single day.
You’ve proposed a three-tier measurement model to help marketers avoid the “false sense of precision” that comes from aggregate scores. Can you walk us through how that works in practice?
The problem with a simple average is that it hides the reality of the market; if you have 90% visibility on Perplexity but only 30% on ChatGPT, an average of 60% tells you absolutely nothing useful if Perplexity only accounts for 1% of your traffic. My Tier One includes platforms with both scale and strategy—ChatGPT, Gemini, and Claude—which should be tracked individually. Tier Two is for embedded AI, specifically Google’s AI Overviews and Microsoft’s Copilot, which reaches 900 million monthly active users across its product family. Tier Three is for the specialists and the emerging players like Perplexity, Grok, or DeepSeek, where you monitor them for unusual growth or specific citations without letting them distort your primary metrics. The key is to connect three data sets: audience exposure, visibility (like mentions and linked URLs), and actual business impact. By de-weighting the smaller players instead of deleting them, you maintain a competitive signal while ensuring your primary focus remains on the platforms that actually drive revenue.
What is your forecast for the AI search market over the next two years?
I expect we will see a “great thinning” of standalone AI assistants as the market converges on platforms that offer deep ecosystem integration or specialized professional utility. Google and Microsoft will likely dominate the general discovery layer, but we will see a surge in “hidden” AI search—instances where people are using tools like Claude or specialized enterprise models to perform deep research that never touches a traditional search engine. We are moving away from a world of “searching” and into a world of “synthesizing,” where the value won’t just be in who appears in a list of links, but who is cited as the authoritative source in an AI-generated brief. Marketers who focus on building a footprint across these different tiers of AI, rather than just chasing the highest referral percentage, will be the ones who survive this transition. The platforms we track will continue to shift, but the fundamental need to be where the conversation is happening remains the only constant in search.
