Gambling SEO After AI Overviews: How Affiliates Get Cited When Fewer Than a Third of Searches Send a Click

68% of U.S. Google searches from January through April 2026 ended without a click. Google never mentions gambling in its AI Overviews guidance document, while public affiliates have pointed to “poor organic search dynamics” behind weaker results two quarters in a row. We’ll break down what’s actually being measured, where AI responses get their sources, and what affiliates should do about it.

Key Takeaways

• The share of Google searches that send a click is shrinking, but the numbers can’t be added together. According to SparkToro, on the SimilarWeb panel, from January through April 2026, 68.01% of Google searches in the U.S. have ended without a click. In 2024, that number was 60.45% (different panel). Ahrefs found top-position CTR 58% below its forecast when an AI Overview appeared (February 2026), versus 34.5% in April 2025; correlation, not causation;
• Google does not name gambling in its AI Overviews documentation. It says only that YMYL queries get an “even higher bar” for supporting sources, and that its goal is to keep AI Overviews off “highly sensitive, explicit or dangerous topics”. There are no additional requirements to appear in AI Overviews, and no separate policy for casino/betting queries was found — frequency must be measured independently;
• And finally, it’s costing money. On October 16, Catena Media warned of a 29% drop in Q3 revenue and a write-down of €30–35 million, citing generative search among several factors (also AI-assisted personalised search, search engines elevating lower-relevancy content, and seasonal summer budget cuts by some operators). At Grandstand (formerly Gambling.com Group), two-thirds of marketing revenue is not driven by SEO. Better Collective, by contrast, grew revenue 9%, driven by North American revenue share, talent-led media, prediction markets and sponsorship.

How Many Clicks Are Left: Six Measurements, Five Methods

The headline figure comes from a study by Rand Fishkin, co-founder of SparkToro (June 9, 2026). He analysed Similarweb clickstream data (desktop and mobile browser, U.S., January–April 2026): 68.01% of Google searches ended without a click. His takeaway is short: “Google is becoming a walled garden.”

In 2024, SparkToro counted 60.45% on a different panel (Datos); Fishkin himself warns that comparing the two is “apples and oranges”. He also pointed out that the share of searches after which the user simply ran a new search rose by 7.2 percentage points. Only 0.34% of searches ended in Google’s AI mode — even though Google announced at I/O 2026 that AI Mode has more than one billion monthly users (a Google figure cited by SparkToro).

Ahrefs measures differently — not people’s behaviour but Google Search Console data. Ryan Law took 300,000 keywords — 150,000 with an AI Overview and 150,000 informational keywords without one — and compared average desktop CTR for position one. April 2025 (March 2024 vs March 2025): CTR fell from 0.073 to 0.026 against an expected 0.040, a gap of 34.5% to the forecast. February 2026 (same method, December 2023 vs December 2025): 0.073 → 0.016 against a forecast of 0.037, a gap of 58%. The effect fades down the page: −50.8% at position two, −19.4% at position ten. Ahrefs stresses this is correlation, not causation.

On July 22, 2025, Pew Research Center published a panel study based on the browser logs of real users: 900 American adults, March 2025, 68,879 unique searches. When an AI summary appeared on the page, users clicked a regular result in 8% of visits, versus 15% when there was no summary. Clicks on a link inside the summary: 1% of visits. 26% of visits with a summary ended the browsing session, versus 16% without. Summaries appeared in 18% of searches.

Saharsh Agarwal (Indian School of Business) and Ananya Sen (Carnegie Mellon) carried out the only randomized experiment on AI Overviews that we found. This involved 1,065 U.S. participants, who over two weeks (January–February 2026) used a Google Chrome extension that was randomly set up to: 1) do nothing; 2) hide AI Overviews; or 3) redirect them to AI Mode. Their plans were pre-registered, and the resulting paper was posted on SSRN on April 3, 2026 (it has not gone through peer review). The authors show that on queries that featured AI Overviews (they did so in 42% of queries), organic clicks fell 38% (SEJ; some reviews give 39.8%), and the proportion of queries that resulted in no clicks increased from 54% to 72%. But when the AI Overviews were hidden, satisfaction with search results was unchanged, leading them to conclude that AI Overviews “divert traffic away from publishers without delivering measurable improvements in user experience.”

Table 1. Actual click metrics following AI Overviews. Sources: 1–5, 11. The studies use different panels, time periods, and definitions of “click”; the figures are not comparable with one another.

Ahrefs and Semrush cover all topics; Pew and Agarwal study the real searches of ordinary people, where betting and casinos are a tiny share. None of them singles out gambling. Anyone talking about affiliates losing X% of traffic because of AIO is either using their own unpublished data or extrapolating from other data.

Does Google Show AI Overviews for Gambling-Related Search Queries?

In short: Google doesn’t write about this. The document “AI Overviews and AI Mode in Search” (May 2025) contains three relevant statements. For YMYL queries — “where information quality is critically important” — the company applies an “even higher bar” for supporting sources. Its systems are set up “with the goal of preventing AI Overviews from appearing on queries about highly sensitive, explicit or dangerous topics”. And AI Overviews “won’t appear on every query” — only where a generative answer “can be especially helpful”. The words “gambling”, “betting” and “casino” do not appear in the document.

The Search Central article “AI Features and Your Website” (updated December 10, 2025) adds two practical points. First: “there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary” — the normal indexing and snippet rules apply, and there’s no need for any special schema.org tagging. Second: Google explains that both features work by “query fan-out” from the original search query to generate additional related queries on various sub-topics. So it’s possible that a website could feature in the answer for a search term even if it isn’t among the top few blue-link search results for that term. And finally, AI Overviews cannot be turned off, and the user guide honestly states that “AI Overviews can and will make mistakes.”

There is an indirect clue in Semrush’s data: among 200,000 keywords that triggered an AIO (September 2024), transactional queries were under 3%, and 80% of desktop AIOs were informational. This suggests that informational guides such as ‘how do wagering requirements work’ or ‘is online gambling legal in Texas’ — the top-of-funnel guides review sites used to win traffic — make up a larger share of AIOs than money keywords such as ‘best casino bonus UK’ or ‘bet365 bonus code’. It does not show how often each type triggers an AIO, and nobody has published a gambling-specific sample with a method. Vendor blogs claiming that Google is “more cautious” with AIO on betting in Europe and Australia provide no data, so we do not use them. The practical takeaway is one: take 30–50 of your own commercial and informational keywords per market and record weekly whether an AIO appears and whom it cites.

Who Do AI Responses Quote?

Peec AI (a GEO analytics vendor) has the largest publicly available dataset: 30 million cited sources from ChatGPT, Google AI Mode, Gemini, Perplexity, and AI Overviews, U.S., March 2026. Top 10: Reddit, YouTube, LinkedIn, Wikipedia, Forbes, G2, Yelp, Facebook, Medium, TechRadar. But even Peec concedes that different industries have different leaders, and, for example, in the hotel sector, Booking.com and Airbnb are ahead of Reddit. So far, nobody has publicly released this breakdown by industry for gambling.

Semrush (another vendor) began publishing an AI Visibility Index in September 2025. The November 4, 2025, report notes Reddit is now the second-most-cited source in Google AI Mode, with its presence up 75%. One interesting note, though, is that ChatGPT and AI Mode overlap on only 67% of brand mentions, and on only 30% of the sources cited. In the June 26, 2026 report, based on 126 million prompts in the United States from January through April 2026, Semrush says that the average AI response from ChatGPT includes 15 citations, heavily dominated by Reddit and Wikipedia. By comparison, Gemini includes an average of only three citations. On Gemini, according to Semrush, the overlap between brands named and sites cited can be as low as 30%, meaning that a brand could be mentioned without citation, or cited without mention.

Two surprising facts about Google itself. It’s worth noting that sources like Wikipedia, Reddit, and even YouTube, combined, account for only 15% of sources in Google AI summaries, versus 17% in traditional results, according to Pew. In other words, forums aren’t heavily represented in AI summaries, contrary to what some can claim. And the top-ranking organic result is not always included: Semrush says that on desktop, the top-ranking organic result is referenced in only 46% of AI answers, and just 34% on mobile. More than half of AI answers don’t reference it. And remember, citations and rankings are two distinct mechanisms — citations are driven by fan-out based on sub-questions.

What Public Affiliates Say

Catena Media (CEO Manuel Stan) published its Q2 results on August 11. The Q2 2026 report was titled “Flat revenue and lower earnings as organic search headwinds persist”, showing revenue of €9.5 million (down 1% YoY), adjusted EBITDA of €1.2 million (down 11% YoY) and 24,781 new depositors (+23%). Stan described “industry-wide headwinds in organic search” but stressed his intention not to abandon organic search or traditional SEO. In Stan’s words: “To be clear, this is not a retreat from organic search and traditional search engine optimisation.”

Q3 was worse. In its preliminary Q3 results published on October 16, Catena reported revenue of €8.3 million versus €11.6 million a year earlier (down 29% YoY), adjusted EBITDA of €0.7 million versus €2.9 million, and a non-cash write-down of intangible assets of €30–35 million, and it dropped its double-digit organic growth target. Catena named several factors: generative search is reducing new traffic across the affiliate industry; its rankings fell because of “AI-assisted personalised search patterns” and search engines elevating lower-relevancy content; and some operators cut marketing budgets during the usual summer slowdown. Because more than 90% of new customers come via CPA, there is little recurring revenue to cushion this. In Stan’s words: “The search landscape is undergoing a rapid and fundamental shift.”

Better Collective, Gentoo Media and Gambling.com Group — three media firms in the gambling space — have reported divergent second-quarter results. Gambling.com Group, which has been operating under the name Grandstand since July 22, 2026, saw revenue fall 5% to $37.8m in its second quarter (reported August 13). Marketing revenue fell 10% to $26.5m on lower organic search revenue, although CEO Kevin McCrystle says non-SEO revenue is now two-thirds of marketing revenue. Meanwhile, cost of sales jumped 119% (to $5.9m), which the company links to diversifying traffic sources. Data revenue was a bright spot, up 12% YoY to $11.2m. In May, it announced a proposed “AI transformation-led” restructure: 25% fewer staff and $13m in annualised savings. The same release blamed “poor organic search dynamics” for a 5% drop in Q1 marketing revenue. Then-CEO Charles Gillespie noted in a March 12, 2026 earnings release that for the first time in the company’s history revenue from sources other than organic search exceeded that generated via SEO.

Better Collective (August 20) posted second-quarter revenue of €89m, up 9% YoY, with EBITDA (pre-special items) of €27m, up 20% YoY. Growth was led by North America (revenue share up 49% to €6 million, alongside talent-led media and prediction markets) and by sponsorship revenue (group-wide +39%, Publishing +44%). Better also runs a publishing arm and noted in its earnings release that its audience “mainly comes from direct traffic and organic search results,” and that it has launched an end-to-end agentic content creation tool. Gentoo Media (August 26), the parent of AskGamblers, Time2Play and CasinoTopsOnline, said its second-quarter revenue of €22.9m was down 9% YoY, and cut its full-year revenue guidance to between €97–100m from its previous view of €105–115m. Gentoo said the softer performance was down to its portfolio rationalisation and weaker sports betting margins, and did not mention Search, Google or AI in its release. Four companies, one market, four different results: one explanation is the portion of revenue that does not depend on a company’s ranking on Google.

Playbook: How to Get Cited—Not Just Clicked On

Below are ten steps. The “Evidence” column contains only what is found in the sources; where our judgment is involved, it is clearly stated as such.

Table 2. Checklist for Gambling Affiliates. Sources: 3, 6, 7, 9–18, 22.

What not to do: don’t clutter every page with “AI-FAQs”. Google explicitly states that no special markup or “special optimizations” are required, spam is handled by core anti-spam systems plus AIO-specific updates, and tries to keep AI Overviews off “data voids” queries (6, 7).

Tracking AI Traffic: What Happened to utm_source=chatgpt.com

For context, OpenAI developers added in June 2025 a utm_source=chatgpt.com URL parameter to outbound links generated within the ChatGPT service. Since ChatGPT usually doesn’t provide a referrer string, this is how most sites differentiate ChatGPT traffic from Direct traffic. As reported by Barry Schwartz on Search Engine Roundtable on October 14, 2026 (attributed to a tip from Chi Lo), after the release of GPT-6 (October 8), the URL parameter and referrer string were suddenly missing, with ChatGPT traffic being reclassified as Direct. According to a demonstration by Adobe’s Cedric Hüsler, this was due to links opening in a new window using window.open with the “noopener,noreferrer” attributes set. Based on a sample of roughly 500 pages, with a ChatGPT share of at least 5% of traffic, over the period of October 9–13, Hüsler estimates a loss of roughly 31,000 ChatGPT visits per day and an increase of roughly 26,000 Direct visits per day.

Hüsler reports that OpenAI confirmed that a bug caused this and that they released a fix that started rolling out around 2:00 a.m. Pacific Time on October 15. By around 9:00 a.m. PT, ChatGPT-referred traffic was back at about 95–100% of what it was the week before. Hüsler’s reaction to this was succinct. “Relieved. It was a bug.” So far, OpenAI has not released an official statement, and this bug fix does not show up in their October 1–15 changelog. As of October 25, LLM Pulse, a vendor, finds the tag in 97.53% of ChatGPT responses that include a link. Traffic from ChatGPT links was mostly renamed during the period and not lost, so annotate October 8–15 in all reports, compare ChatGPT and Direct page by page, and do not rebuild attribution because of the dip.

What This Means

To affiliates. Top result CTR impact studies vary considerably (Ahrefs sees 34.5–58% below the forecast, Pew sees 8% versus 15% clicks, another experiment reported a −38% dip in organic click-throughs) — an indicative range from non-comparable measurements, not a single figure. The impact on your business will become clearer as you monitor AIO on your own queries. If your business heavily depends on informational rather than transactional searches, it may be worth examining the breakdown of revenue by page, and how many pages are directly answered by Google.

To operators. For those relying heavily on affiliates that source most traffic through organic search: Catena suffered a 29% drop in Q3, after a flat Q2. Affiliates that acquire more than 90% of new customers under CPA deals (as Catena does) may experience higher volatility in 2026. As part of due diligence, be sure to ask affiliates about their revenue source breakdown (brand/direct traffic ratio, for example), as well as the proportion of revenue not earned through organic search, which for Grandstand is two-thirds of marketing revenue — it is now a key metric for stability.

To SEO teams. Track rankings and clicks separately from citations and brand demand, and AI referrals with an October annotation, as there are no technical requirements for inclusion in an AI overview (according to Google). However, keep in mind the fan-out effect by sub-queries, the likelihood of citations from pages other than your top-ranking organic pages, and the relevance of Reddit given its role in AI Mode (Semrush, autumn 2025). Annotate October in AI-referral reports.

Methodology

These statistics are current as of November 4, 2026. Table 1 includes only studies with a published sample and method. We have not included data from statistics aggregators, and we have not compared findings from different authors. For example, SparkToro reports the proportion of searches that result in no click from their own panel; Ahrefs reports click-through rates for the top result based on data from Google Search Console; and Pew reports clicks based on visits from its own panel. The study by Agarwal and Sen involved a controlled experiment in which participants installed a browser extension. Data from Semrush, Peec, Ahrefs, and LLM Pulse are vendor data: all four companies sell analytics and other tools that provide search information and information about the visibility of AI. Catena Media’s Q3 numbers are preliminary, and the final report will be released on November 10, 2026. All numbers are in each company’s reporting currency. Direct quotes are given in the original wording.

Sources

1. SparkToro (Rand Fishkin), “In 2026, Less than One Third of Google Searches Still Send a Click”, 09.06.2026 — 68.01%, 60.45% in 2024, 0.34% AI Mode, +7.2 percentage points in repeat searches. https://sparktoro.com/blog/in-2026-less-than-one-third-of-google-searches-still-send-a-click/
2. Ahrefs (Ryan Law), “AI Overviews Reduce Clicks by 34.5%”, April 17, 2025 — 300,000 keys, 0.073 → 0.026. https://ahrefs.com/blog/ai-overviews-reduce-clicks/
3. Ahrefs (Ryan Law), “AI Overviews Reduce Clicks by 58%” (research update), February 4, 2026 — 0.073 → 0.016; −50.8% and −19.4%. https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/