Of roughly one million high-volume keywords tracked through 2023, 40.7% lost more than 15% of their volume year over year. The average drop among those affected keywords was 41%. That's not a rounding error. That's a structural shift in where demand gen teams can expect top-of-funnel traffic to originate.
But total search volume across the same dataset? Flat. Gains offset losses almost perfectly: ~10.31 billion searches gained per month versus ~10.29 billion lost. Net change: +16.8 million. Search isn't dying. It's redistributing.
The queries AI eats first
Information-heavy categories took the hardest hit. Drug interactions, insurance deductibles, fund overviews, product definitions. Anything where a chatbot can deliver a complete answer in one shot without the user needing to click through to a site. FinTech keywords dropped 37.7%. Lifestyle held up better at -15.2%.
If your content portfolio leans on informational queries (and most B2B SaaS programs do, because that's where volume lives), you're exposed. Help docs, glossary pages, "what is X" articles, comparison tables built around basic feature lists: these are the categories getting absorbed by AI-generated answers before the user ever sees a SERP.
AI Overviews now appear in 99.9% of informational keywords. Nearly half of those are long-tail (7+ words), and 57.9% are question-based. When an Overview shows up, organic CTR drops 61%. About 60% of traditional search queries now end without a click, thanks to AI summaries. Zero-click rates jump to roughly 26% with AI summaries present, compared to 16% for traditional results.
The citation game changes the math
Here's where it gets interesting for operators. When a brand gets cited inside the AI Overview, CTR increases 35%. And 59% of consumers say they're likely to visit a brand's website after an AI chatbot mentions or recommends it. Gen Z and millennials are 2.5x more likely than boomers to buy based on an AI recommendation (20% vs 7%), even unverified ones.
So the traffic isn't gone. The funnel entry point shifted. Instead of ranking on page one and earning a click, the new leading indicator is whether your content gets referenced by the AI that sits above page one.
Experts across the industry describe this as a move from "ranking for search" to "becoming the source AI cites." The signals that earn citations look different from classic ranking factors: content clarity, trustworthiness, specificity, original research, recency, and being frequently referenced by other authoritative sources. First-hand experience and proprietary data carry weight that generic SEO content doesn't.
What to actually change in your program
Traditional keyword research is becoming insufficient on its own. The recommendation gaining traction is what some call "Prompt Research": figuring out how your audience asks AI tools for advice and which brands surface in those answers. That's a different input than search volume and keyword difficulty.
Dual optimization is the operating model. You optimize for classic search (where clicks still happen, especially on high-intent queries) and for generative AI experiences (where citations drive awareness and downstream visits). Structured content matters for both: chunking, content clusters, pillar pages, schema clarity. These aren't new tactics, but the reason to invest in them has a new edge.
A few concrete moves worth running:
- Audit your keyword portfolio for AI Overview exposure. Which of your top-traffic pages target informational queries where Overviews appear? Those pages need a citation strategy, not just a ranking strategy.
- Map your AI citation presence. Ask ChatGPT, Perplexity, and Gemini the questions your buyers ask. Note which brands get cited. If you're absent, that's the gap.
- Shift content investment toward original research and experience-based assets. AI-written pages show up in 17% of top results already, and they're vulnerable to algorithm changes. Differentiated content (proprietary data, real case studies, named expert perspectives) is harder to replace and more likely to get cited.
- Build new measurement. AI referrals are hard to attribute (ChatGPT, for instance, sends limited UTM data). Define new KPIs: AI Overview citation rate, zero-click brand exposure, assisted conversions from AI mentions. Marketing Ops needs to own this.
The trade-off nobody wants to hear
Experts project 15–25% of organic search traffic could migrate from Google to LLMs within 24 months. At the same time, traditional SEO's share of total website traffic is projected to rise from 45% in 2025 to 53% in 2026. Both things can be true because the pie is growing and reshaping simultaneously.
The risk isn't that SEO stops working. The risk is that your SEO program is optimized for a query mix that's already shifting underneath you. Volume on resilient, high-intent keywords may actually increase as informational queries drain away to chatbots.
Twelve months ago, the question was whether AI search would matter. The million-keyword dataset answered that. The question now is whether your content earns the citation or just the ranking. One of those still drives pipeline. The other is starting to drive impressions nobody clicks.
