Two weeks ago at Google’s I/O conference, the company described its new search experience as “AI search through and through.” The ten blue links — the search results page (SERP) links everyone fights over— will soon go away, and they’re increasingly not the point. More and more, Google is answering questions directly and not sending users to websites. According to recent data, 58% of Google searches now end without a single click.
I’ve waited to post this so that I could read about three dozen articles about how this will affect all of us. Jay Peters wrote a great piece on The Verge about it (note paywall)
I once made my living pushing content for clients to get them to the first page of SERPs for their keywords, sometimes just to be up top somewhere on Google, and sometimes for reputational management. It was mostly white-hat stuff, some gray-hat stuff. But it was simple: do good work, publish clear content, and Google will send people your way. The ten blue links essentially served as a free distribution channel. You earned your spot through relevance and consistency.
That model is going away. For nonprofits that built their web presence around organic search, this is a real problem. And most haven’t started dealing with it yet.
What’s actually happening
You may have heard the term “AI Overviews” — those AI-generated summaries that now appear above search results for a huge range of queries. Google rolled them out broadly in 2024, and by early 2025, they were showing up in more than half of all searches. The click-through rate impact has been significant: independent research found drops of 34–46% on queries where AI summaries appear. Some publishers have reported losses of80–90% for certain content types.
The I/O 2026 announcement takes this further. The new “AI Mode” isn’t just a summary at the top of the results page. It’s a full conversational interface — you can ask complex questions, get synthesized answers with charts and data, follow up, and go deeper, all without ever leaving Google. A user can now ask “What’s the most effective way to help homeless youth in my city?” and get a thorough, sourced answer. Your nonprofit might be part of that answer. Or you might not. Either way, they probably didn’t visit your website.
Google lead product manager Logan Kilpatrick has strongly hinted that AI Mode will eventually replace traditional search as the default experience. Lily Ray, VP of SEO strategy at Amsive, says that the changes would “severely cut into the main source of revenue for most publishers” and disincentivize content creators who rely on organic search traffic.
Why nonprofits are especially exposed
For-profit companies have other ways to reach people — paid marketing, brand recognition, and online commerce. Nonprofits, especially smaller and mid-size orgs, lean heavily on organic search because it’s free and generally works. A lot of that content — the informational, explainer-type pages that answered questions like “what is food insecurity” or “how does affordable housing financing work” — is exactly the kind of content AI overviews answer directly and completely.
The Reuters Institute for the Study of Journalism says search engine referrals will fall by 43% over the next three years.
There’s also a Google Ad Grant angle here. The $10,000/month grant is valuable, but paid results also take a significant hit when AI Overviews appear — click-through rates on paid links drop roughly in half in those contexts. The grant doesn’t lose its value, but it needs to be paired with a smarter content strategy to do its job.
There’s an upside
Here’s the thing that doesn’t get said enough in these conversations: the visitors who do click through from AI-cited sources tend to be higher-intent than traditional search visitors. Early data suggests they’ve already asked their questions and gotten context. When they land on your site, they’re not browsing around — they’re ready to act.
Being cited by AI is the new version of ranking on page one. And the criteria for getting cited — clarity, authority, depth, trustworthiness — are things good content teams already know how to do. It’s not a completely different game. It’s the same game with different rules.
What to actually do
Invest harder in channels you own. Rely on things you probably already have: newsletter and alert lists, text, partners, and audience help with social channels and your donors. Search was always a rented audience — you just didn’t always feel the rent going up. Now you do. The Mangrove Web team frames it clearly: orgs that built direct lines to their audiences are far better positioned for this shift than those that relied on search discovery alone.
Think about being cited, not just ranked. AI systems don’t simply cite the top-ranked page. SEMRush suggests only about 40–50% overlap between what AI Mode cites and what shows up in traditional top-10 results. The rest is selected based on perceived trust, clarity, and how well the content answers the full question. This shift — from SEO to what practitioners are now calling Answer Engine Optimization (AEO) or Generative Engine Optimization (GEO) — changes how you should think about content investment.
Go deep on fewer pages. The old playbook said to write a separate page for every keyword. AI systems prefer comprehensive, authoritative content over a scattered collection of thin pages. A single well-structured page that covers your program model, your evidence base, your impact data, and how people can get involved is more likely to be cited than five separate pages that each cover a piece of the story. Good Dog Strategies puts it well: AI models aim to find a “pillar page” that covers the user’s entire thought process.
Make your expertise explicit. Named authors. Credentials. Links to primary data. Specific outcomes with real numbers. Vague mission statements don’t get cited. Specific, verifiable information does. This connects directly to what I wrote last month about structuring your content for LLM readability — the same signals that help AI systems trust you apply here.
Check what’s happening in your data. Use Google Search Console data to examine year-over-year organic traffic trends for your informational content. If things are down, you’re not alone, and you’re not doing anything wrong — but it’s better to know and respond strategically than to keep optimizing for a world that’s shifted.
This is the same shift I wrote about last month from the LLM side — the question of whether AI systems know who you are and trust what you say. Google’s AI Mode is another version of that same question, just asked at the front door of the internet.
The blue links are still there. But they’re not running the show anymore.