Milan Bogojevic Blog

AI Overviews and Nonprofit Traffic: What the 2026 Data Shows

8 min read updated 10 October 2026

Split-screen Google search: blue links on the left, an AI Overview dominating the right
Split-screen Google search: blue links on the left, an AI Overview dominating the right

AI Overviews are associated with fewer organic clicks in several major studies, but their impact varies by query, period and whether a brand is cited. A nonprofit should track impressions, qualified visits and completed actions separately. The 2026 evidence supports watching citation alongside ranking; it does not justify applying one sector-wide loss percentage to your own website.

Why this matters for a small comms team

Most nonprofit communications teams don't have a data analyst watching search console dashboards. They notice a traffic drop, assume something broke, and spend time debugging a website issue that doesn't exist. The real cause is often structural: the query that used to send a visitor now gets answered directly on the results page, by Google, before anyone reaches your site at all.

This matters more for programme and donor pages than for blog content, because those are the pages donors and beneficiaries use to decide whether to act (donate, apply, volunteer), and a silent drop in their traffic is easy to miss until the quarterly numbers already reflect it.

The nonprofit reality: a disclosed, hypothetical example

Imagine a nonprofit running a well-ranked page titled "How to apply for emergency housing assistance." For two years it reliably sent visitors to the application form. In 2026, someone searching that exact phrase increasingly gets a direct AI-generated answer on the results page itself, often summarising eligibility rules without ever naming the organisation, let alone linking to the application form.

The page didn't get worse. The way people reach it changed. Fixing this isn't a ranking problem, it's a citation problem: making sure that when the AI Overview does summarise eligibility, it names this organisation specifically and links to the actual form, rather than paraphrasing generic criteria from elsewhere.

The Citation Response Loop

Once a team accepts that ranking alone won't bring back lost clicks, there's a simple, repeatable response. I call it the Citation Response Loop, because it's meant to run monthly, not once.

Measure → Target → Strengthen → Monitor

  • Measure: Record AIO presence for a small query panel alongside page-level clicks and conversions. Standard analytics does not identify every visitor's query or whether they saw an AIO, so label this as an observational comparison.
  • Target: Rewrite your most important pages so the direct answer sits in the first two or three sentences, since that's what gets lifted into a citation.
  • Strengthen: Add named authors, real credentials, and attributed expert quotes to the pages you most need cited, since trust signals matter more when AI has to choose a source to name.
  • Monitor: Use the AI visibility audit and check whether AI search bots (OAI-SearchBot, Claude-SearchBot, PerplexityBot) can actually reach your priority pages, separately from training-only bots like GPTBot or ClaudeBot.

What do the traffic studies actually measure?

These studies use different queries, samples and periods. Read each result within its own design; none gives a universal nonprofit traffic-loss forecast.

Study Verified finding Limit for a nonprofit team
Seer, September 2025 update Organic CTR on its AIO-query cohort fell from 1.76% in June 2024 to 0.61% in September 2025. It tracked 3,119 terms across 42 organisations. The AIO cohort already had lower CTR at baseline; this is not a clean before/after causal estimate.
Seer, April 2026 update The expanded study covered 53 brands, 5.47 million queries and 2.43 billion organic impressions, with actuals through February 2026. Some early-2026 segments stabilised or recovered. A larger, changed cohort cannot be appended to the older figures as one continuous recovery series. March 2026 figures were projections.
Ahrefs, February 2026 Across 300,000 keywords, December 2025 data associated AIO presence with 58% lower top-result CTR relative to the study's forecast. This compares observational keyword cohorts; it is not a prediction of a 58% loss for your whole site.
Pew, July 2025 In 68,879 unique Google searches, traditional-result links were clicked on 8% of visits with an AI summary versus 15% without one. It measures user visits in a US browsing panel, not the same CTR definition as Seer or Ahrefs.

The practical question is whether your priority pages are losing qualified visits, and whether people can still reach the right service or donation action.

Seer's 2025 update reported 35% higher organic CTR and 91% higher paid CTR for cited brands in its Q3 comparison. These are relative CTR differences, not proof that earning a citation causes the improvement. The 2026 update found a different citation premium in its expanded dataset: citation remains worth measuring, but its effect is not a fixed percentage.

Ahrefs' March 2026 citation study found about 38% of AIO citations came from the original query's top ten results, compared with roughly 76% in its July 2025 study. Ahrefs also changed its parsing method, so the difference is not a clean measure of one algorithm change. BrightEdge's February 2025-February 2026 analysis reported about 17% top-ten overlap using its own dataset. Do not combine those figures into one trend.

Ranking and citation are connected, but neither guarantees the other. Keep measuring qualified clicks and completed actions alongside mentions.

What still works, and what's new

Google has stated no special schema markup is required for AI Overview citation, and I treat structured data the way Google itself frames it: a way to remove ambiguity for a crawler, cheap and plausible to add, not a lever that pulls a citation on its own. Article and Organisation markup are worth adding if they match your visible content exactly; FAQPage markup only belongs on a page with a real, visible FAQ.

If your About page is the place this markup points to, it's worth checking that the page itself actually answers the questions a stranger would ask, covered in Your About Page Is Doing More Work Than Your Homepage.

Google's AI-features guidance says standard search requirements and policies still apply. Aggarwal et al., KDD 2024 tested citations, quotations and statistics in generative-engine responses and found visibility gains in that experimental setting. This is not a Google ranking rule or a promised increase in nonprofit clicks.

My view

I'd push back on comparing this moment to "the biggest shift since the mobile index rollout," which is how this topic usually gets framed. Mobile indexing changed how pages were crawled and ranked. This changes whether a ranked page gets visited at all. My own rule for a nonprofit client is blunt: if a page's whole job is to drive an action (apply, donate, volunteer), I now treat its AI Overview citation rate as more important than its ranking position, because a page can rank first and still send nobody.

You cannot measure the second channel completely yet. Referral data from AI answers remains thin and inconsistent across providers. A GEO dashboard measures a proxy: write for extraction because it is a reasonable bet about how people find information, while evaluating actual enquiries and completed actions separately.

What to do next

  • Pull your analytics for your five highest-traffic programme or donor pages and check whether their click trend changed around when an AI Overview started appearing for their main query.
  • Rewrite the opening two to three sentences of each of those pages so the direct answer comes first, before any context or background.
  • Add a named staff author and a one-line credential to any page where that's missing.
  • Check your server logs (or ask whoever manages your site to check) for OAI-SearchBot, Claude-SearchBot, and PerplexityBot activity on those same pages.
  • Use the 30-day sub-query plan to close useful content gaps, and repeat a small prompt panel monthly.

Conclusion

The data through early 2026 is consistent on one point: a page that ranks well but isn't cited inside an AI Overview is increasingly likely to get seen and not clicked. That's a harder problem to notice than a ranking drop, because the traffic decline looks gradual and the page itself hasn't visibly changed. The response isn't to abandon SEO fundamentals, rankings, speed, and backlinks still matter as prerequisites, but to add citation rate as a metric a small team actually tracks, and to treat it as the number that predicts whether anyone shows up.

FAQ

Does this mean SEO is dead for nonprofits?

No. Search fundamentals still matter. Ranking and AI citation overlap, but neither guarantees the other. Keep tracking qualified search visits while measuring AI mentions separately.

Is schema markup required to get cited in an AI Overview?

No. Google has stated no special schema is required. Structured data like Organisation and Article markup helps remove ambiguity for a crawler, but it doesn't function as a citation lever by itself.

How much organic traffic should a nonprofit expect to lose to AI Overviews?

There is no reliable universal percentage. The cited studies use different samples and definitions. Compare your own priority pages over time, record AIO presence, and check completed applications, donations or enquiries before changing your strategy.

Do we need an expensive analytics tool to track this?

Not necessarily. Splitting your existing click and conversion reporting by whether an AI Overview appeared for a given query is the core step, and that can often be done with a manual check on your highest-priority pages before investing in a dedicated tool.

Sources

About the author

Milan Bogojevic works on nonprofit visibility across search and AI, with a focus on clear content, organisational identity and verifiable evidence.

Last updated: 10 October 2026.

Next in AI SEO & GEO