Gen AI website traffic share update - August 2026 🗓️ 12 months ago: ChatGPT: 73.3% Gemini: 12.9% DeepSeek: 4.0% Grok: 2.6% Copilot: 2.0% Perplexity: 2.0% Claude: 1.9% 🗓️ 6 months ago: ChatGPT: 56.7% Gemini: 25.4% Claude: 6.0% Grok: 3.7% DeepSeek: 3.4% Copilot: 1.9% Perplexity: 1.6% 🗓️ 3 months ago: ChatGPT: 52.7% Gemini: 27.8% Claude: 9.2% DeepSeek: 3.6% Grok: 2.5% Copilot: 2.0% Perplexity: 1.1% 🗓️ 2 months ago: ChatGPT: 54.2% Gemini: 26.5% Claude: 9.5% DeepSeek: 3.4% Grok: 2.5% Copilot: 1.6% Perplexity: 1.0% 🗓️ 1 month ago: ChatGPT: 55.5% Gemini: 25.6% Claude: 9.3% DeepSeek: 3.4% Grok: 2.4% Copilot: 1.6% Perplexity: 0.9%

What changed after the August 2026 Spam Update? To answer that, we first had to establish what changes on Google when there is no update. In a quiet period with no confirmed update, 9.2% of URLs in Google's top 10 drop out of the top 100 completely. That's baseline churn. During the August update, that figure reached 16.71%. That means around 80% more top-10 URLs disappeared from the top 100 than during a quiet period. The change was concentrated in losses, not in new pages breaking through. Entries into the top 3 from outside the top 20 barely moved: 12.86% against an 11.5% baseline. And Reddit? Almost untouched. Its share of top-3 rankings has remained remarkably stable across the last three Google updates. While people are reporting a decline in Reddit citations in ChatGPT, its presence in Google’s top results appears unaffected. For more data, check out the graphs:

AI Overviews have just landed in France, and here’s everything you need to know:

AI Overviews are moving deeper into commercial search. In our six-month study, their presence on commercial-intent SERPs grew 71%, while transactional-intent SERPs declined 5%. We analyzed 600,000+ U.S. keywords across 10 industries from November 2025 to April 2026, measuring AI Overview presence by search intent, industry, CPC, and overlap with Google Ads. 📈 Finance commercial queries: +231% 💰 AI Overview + Google Ads: appearing together ~2x more often than a year ago 🔎 AI Overview SERPs: higher average CPC across most industries AI optimization is moving deeper into the commercial journey. Brands need to identify which topics and products are triggering AI Overviews, connect those opportunities to their content strategy, and choose tools that can track visibility across the searches that matter most. Full study: social.semrush.com/4gPkBEM.

Exa is the default web search for agents built with @Vercel's Eve. Ask your agent anything that needs the web and Eve searches with Exa automatically. Search built for AI, now built into Eve.

Your website still matters. It’s just no longer the only visibility layer. AI tools cross-check everything – Reddit threads, LinkedIn posts, reviews, third-party mentions – to decide what’s credible. Semrush One tracks SERPs and AI citations, so you see the full picture. social.semrush.com/44Aho5n.

Marketing changes fast. What worked last year may not work as well this year. Just look at what channels provided growth in 2025 versus what's working today.

Our latest "Downstream Impact of AI Visibility" report shows: → AI tools are most useful at the start of the consumer journey. → During the research and comparison stage, AI still leads. → When users are ready to find where to buy, traditional search catches up. >>

Google May 2026 core update volatility spikes big time today, Tuesday, June 2nd - just when you thought it was over... seroundtable.com/google-may-…

Can scaling AI content be risky for SEO? I've been monitoring the impact across hundreds of sites for the past few months. Check out my recent research and findings in my latest Substack: lilyraynyc.substack.com/p/it…

Here's how much AI traffic is driven by industry. Conductor did an analysis of 3.3 billion sessions and found that 1.08% of website traffic comes from LLMs. Over time, you should see this percentage increase. And although it's not a big number, keep in mind that AI referral traffic is the highest-converting traffic we have ever seen.

Not all sites with self-promotional listicles have been hit, but for the handful that were hit on ~Jan 21, the crash is still crashing. As a reminder, these sites used dozens/hundreds/thousands of self-promotional listicles (plus other spammy approaches)

In the coming weeks, we plan to start testing ads in ChatGPT free and Go tiers. We’re sharing our principles early on how we’ll approach ads–guided by putting user trust and transparency first as we work to make AI accessible to everyone. What matters most: - Responses in ChatGPT will not be influenced by ads. - Ads are always separate and clearly labeled. - Your conversations are private from advertisers. - Plus, Pro, Business, and Enterprise tiers will not have ads.

What are you planning to do with your social ad budgets in 2026? Most companies plan to increase it. Here's what networks they are planning to ramp up spending on.

Surfer just published a study that should worry anyone tracking AI search visibility. They ran 2,000 tests comparing API-based data vs. scraped results from ChatGPT and Perplexity. The findings? API results and what users actually see are completely different. Most AI visibility tools use APIs to pull data from ChatGPT and Perplexity. Why? It's easier and cheaper than scraping actual results. The problem? APIs don't show what real users see. Surfer tested 1,000 prompts on ChatGPT and another 1,000 on Perplexity to find out how different they really are. For ChatGPT, the differences were massive: - Length: API responses averaged 406 words. Scraped results averaged 743 words. - Web search: 23% of API responses didn't trigger web search at all. Scraped results always did. - Sources: APIs provided zero sources 25% of the time. Scraped answers always included sources (averaging 16 vs. 7). - Brand detection: APIs failed to detect any brands 8% of the time. Scraped answers always identified brands. But here's the real kicker: Only 24% of brands overlapped between API and scraped results. For sources, the overlap was just 4%. Let that sink in. If you're using API-based tools to track your ChatGPT visibility, you're looking at completely different data than what users actually see. Perplexity showed similar problems: - Length: API responses averaged 332 words vs. 433 words for scraped results. - Sources: APIs returned roughly 7 sources. Scraped results always included 10. - Brand mentions: API responses included 10+ brands on average. Scraped results averaged around 6. - Source overlap: Just 8% between API and real user results. According to Wojciech Korczyński, Data Scientist at Surfer: "These differences are so explicit that monitoring responses from API as a proxy for your AI visibility is totally wrong." If your tracking data is wrong, your optimization strategy will be wrong too. You'll assume certain sources matter when they don't appear in real ChatGPT results at all. You'll think your brand is showing up when it isn't. You'll optimize for the wrong thing entirely. So what should you do? Use tools that rely on scraped data, not APIs. Scraped data shows what real users actually see in ChatGPT, Perplexity, and other AI platforms. It's the only way to get accurate visibility data. Credit to Surfer and Jakub Sadowski for running this study.

NEW: Ramp AI Index shows AI adoption held flat at 45% of businesses in November, driven by slight declines in the finance and technology sectors @tryramp By the model... OpenAI: -1% Anthropic: +0.8% Google: +0.7%

Different AI platforms, totally different citation behavior. And honestly, the gap between them is getting wider every month. Ask the same question across AI platforms and you’ll get three completely different “source universes”. ChatGPT pulls from one mix, Perplexity leans into another, and AI Mode has its own vibe entirely. For marketers, that means your domain can look super visible on one model and almost invisible on another. The tricky part is, most people still assume all LLMs behave the same… and that blind spot can cost visibility fast. 👀 So we compared all 3 side by side and mapped exactly where the citations go across 230K prompts and 100M references. Spoiler: they do not match. Full breakdown, charts and insights here 👉 social.semrush.com/4pGB67Y.