Between spring 2025 and spring 2026, an Irish agency pulled 175 million searches out of Google Search Console and compared the same months year on year. Aggregate click-through rate fell from 3.62% to 2.67%, a drop of 26%. Average ranking position, over the same period, went the other way: it improved slightly, from 6.38 to 6.18. Two metrics that are supposed to move together came apart.
That is the single most useful piece of evidence anyone has produced about Irish search, and it is also the most awkward, because it breaks the reporting habit almost every agency in the country still relies on. If your position improved and your clicks fell, a ranking report tells your board that things got better. The revenue line tells them otherwise.
What follows is an attempt to put that Irish finding next to the four other studies that measure something similar, and to be honest about which of them actually apply here. They do not agree. One of them points in the opposite direction entirely. That disagreement is the interesting part.
What Actually Happened to Irish Search in 2025
Google announced AI Overviews for nine additional European markets on 25 March 2025, with rollout the following day. Ireland was in that group, alongside Austria, Belgium, Germany, Italy, Poland, Portugal, Spain and Switzerland. Irish users got the feature in English only, and at launch it was limited to signed-in users aged 18 and over.
Ireland was not a laggard. France was excluded from that wave altogether, because of its stricter neighbouring-rights regime, and stayed without AI Overviews well into 2026. The rollout also landed one week after the European Commission issued preliminary findings that Alphabet may have breached the Digital Markets Act. So the feature arrived in Ireland early, and it arrived into an already contested regulatory environment.
Then something counterintuitive happened. Despite being an early market, Ireland ended up with comparatively low AI Overview coverage. Ahrefs analysed 108 million AI Overview queries across 50 countries in November 2025 and found the feature on 12.00% of tracked Irish queries. That put Ireland 44th out of 50, behind Portugal, Denmark, the Czech Republic and New Zealand.
Sit with that gap for a moment. If an agency quotes you the American figure of 20.50%, or even the British 19.10%, it is overstating your exposure by something like 60 to 70 percent. That is not a rounding error. It changes how much of your budget should move.
Why the Gap Exists
Nobody has published a causal explanation for Ireland's position in that table, and I am not going to invent one. What can be said is that coverage rates track query composition, and query composition varies by market. Ahrefs' own supporting analysis found AI Overviews appear more frequently on informational queries, longer queries and higher-volume queries, and appear less frequently on branded, local and short ones. A market whose tracked keyword set skews toward branded and local search will show lower coverage. Whether that is what is happening in Ireland is an open question, not a finding.
The Irish Click Study, and What It Does Not Prove
The Friday.ie analysis, published on 6 July 2026, remains the only click-behaviour study conducted on Irish data that I have been able to locate. Its scale is respectable. Its design has one significant limitation, and the authors say so themselves.
The dataset begins in February 2025. AI Overviews reached Ireland in March 2025. So the study has no clean pre-AI Overview baseline. It compares two periods that are both, to different degrees, inside the new regime. That does not make the 26% figure wrong. It does mean you cannot attribute the whole decline to AI Overviews, and any agency presenting it as a clean causal measurement is overreaching.
The more valuable finding in that study is not the headline at all. It is the breakdown by query type.
| Query type | Click-through rate at positions 4 to 10 | What this implies for budget |
|---|---|---|
| Commercial investigation Terms containing words like best, review, compare, pricing | 4.24% | Holding up. Rankings outside the top three still convert attention into visits. This is where organic investment still behaves the way it used to. |
| Informational Definitional and how-to phrasing | 1.16% | Under heavy pressure. A position-six ranking here delivers roughly a quarter of the clicks the same position delivers on a commercial term. |
A ratio of roughly 3.7 to 1 between two query classes at the same ranking position is a bigger practical finding than the aggregate 26%. It tells you that the decline is not spread evenly across your keyword set. Some of your terms are almost unaffected. Others have quietly stopped working. Averaging them together hides both.
And it lines up with the Ahrefs trigger-pattern finding from the other direction. Informational queries trigger AI Overviews more often. Informational queries lose the most clicks. Two independent datasets, same conclusion, arrived at from opposite ends. That is about as much corroboration as this field currently offers.
Four Studies, Four Different Answers
Here is where the field gets uncomfortable. Beyond the Irish study, there are three widely cited measurements of what AI answers do to clicks. They do not agree with each other, and one of them says the opposite of the rest.
The Semrush result deserves more attention than it gets. It is the finding that should stop anyone from treating this as settled. If two credible vendors measure zero-click behaviour over overlapping periods and one says it rose eighteen percentage points while the other says it fell two, then at least one of them is measuring something other than what the headline implies. Most likely both are, and the definitions differ.
What "Zero-Click" Even Means
The phrase gets used for at least three different things. A session that ends without any click at all. A search where the user clicks something that is not an organic result, such as a Knowledge Panel or a Google-owned property. A query whose SERP contains no organic link above the fold. Those produce very different numbers from the same underlying behaviour.
The Agarwal and Sen study is the cleanest of the four because it is a randomised experiment on instrumented browsers, so it observes actual sessions rather than inferring them. It is also American, desktop-only, and ran over roughly six weeks with about a thousand participants. Applying its 39.8% to an Irish B2B client would be a category error, and a fairly serious one, given Ireland's AI Overview coverage sits at 12.00% against the American 20.50%.
A Number That Has Been Widely Misattributed
One specific error is worth flagging, because I have now seen it in three separate places. The 34.5% figure belongs to Ahrefs and measures click-through rate. It does not belong to Agarwal and Sen, and it does not measure zero-click share. Their zero-click finding is the move from 54% to 72%.
The two get fused because they are similar in magnitude and appear in the same paragraph of a lot of secondary coverage. If you see a single study credited with both a 39.8% click decline and a 34.5% zero-click increase, you are looking at two studies stitched together. Check the primary source before that number reaches a client deck.
What This Means for Reporting
The practical consequence of the Irish finding is narrow and specific: ranking position and commercial outcome have partially decoupled, so a report built on position alone can be simultaneously accurate and misleading.
That does not mean rankings are worthless. Position still governs whether you are in the candidate set at all. It means position is now an input metric rather than an outcome metric, and reporting has to separate the two.
A defensible Irish reporting stack looks roughly like this.
| Layer | What it measures | Why it belongs in a separate section |
|---|---|---|
| Position and visibility | Ranking by query class, share of visible SERP real estate, AI Overview presence flag per query | Diagnostic. Tells you whether you are eligible, not whether you benefited. |
| Click behaviour | Impressions, clicks and CTR split by query class, not aggregated | The Irish data shows a 3.7 to 1 spread between commercial and informational terms at the same position. Aggregating destroys the signal. |
| Downstream commercial | Qualified sessions, form completions, pipeline contribution | The only layer your finance function recognises, and the one least affected by measurement fashion. |
| Citation and mention | Whether a brand is named inside a generated answer, with or without a link | Runs on a different retrieval system entirely. Cannot be inferred from any of the three layers above. |
The fourth layer is the one most Irish reporting still omits, and it is the one the aggregate CTR decline should push you toward. If a growing share of queries resolves without a click, then a growing share of your visibility produces no measurable session at all. You either measure the unclicked mention or you accept a blind spot that widens every quarter. We cover the mechanics of that in our work on generative engine optimisation, and the market-specific version in our GEO service for Ireland.
Why Irish Numbers and British Numbers Should Not Be Merged
Most cross-border reporting into Ireland arrives with British benchmarks attached, because the language is shared and the agency relationships often are too. On this particular question that shortcut is measurably wrong.
Britain sits at 19.10% AI Overview coverage. Ireland sits at 12.00%. Same language, same search engine, coverage rates seven percentage points apart. Any target, forecast or budget reallocation built on the British figure is calibrated to a different exposure level.
There is a second reason, and it has nothing to do with search behaviour. The institutions differ. An Irish reader looking at a financial services page needs the Central Bank of Ireland, not the FCA. A data question resolves to the Data Protection Commission, not the ICO. Company filings sit with the Companies Registration Office, tax with Revenue, addresses use Eircodes. Spelling follows British convention in Irish English, so spelling proves nothing either way. Jurisdictional accuracy is the signal that actually differentiates, and it is invisible to a spellchecker.
The Republic and Northern Ireland are also two markets, not one. Different currency, tax regime, regulators and public institutions. Merging them in a keyword set blends two buyer intents and produces recommendations that are subtly wrong for both. That distinction is handled properly in our SEO service for Ireland, and it is the single most common structural error we see in inherited accounts.
Where the Money Is Going Instead
One more data point makes the reporting argument concrete. Irish digital adspend reached EUR 1.146 billion in 2025, up 8% year on year, according to the IAB Ireland and IRM Online Adspend Study published on 14 May 2026. Within that total, paid search took EUR 345 million and grew 4%.
Four percent, against 8% for the market as a whole, 12% for social and 14% for video. Search was the slowest-growing paid channel in Ireland that year. Whatever is happening to the click, the money is already reflecting it.
A caveat on that figure that almost nobody quotes: Google, Meta, Twitter, Snapchat, Pinterest and LinkedIn do not submit data to the study. Their spend is estimated through the researcher's econometric modelling. So the headline total is part census, part model. Worth knowing before you build a board slide on it.
Instrumenting an Irish Account Properly
Concretely, then. Six things worth doing, in roughly this order.
Segment the keyword set by query class before anything else. Commercial investigation, informational, navigational, transactional and comparison. The Irish data shows a 3.7 to 1 CTR spread between two of those classes at identical positions. Until you split them, every aggregate you report is an average across populations that behave differently.
Log AI Overview presence per query, per check. A boolean flag, dated. Coverage is not static and 12.00% is a national average, not your average. Your own category could be well above or below it, and you cannot know without recording it.
Stop reporting aggregate CTR as a headline. Report it by class. If aggregate CTR falls while your commercial-term CTR holds, that is a mix shift, not a deterioration, and it needs a different response.
Set the baseline before you intervene. This sounds obvious. The Irish study itself is a cautionary example of what happens when the measurement starts after the change: you get a real number that cannot be attributed cleanly.
Change one thing at a time. If you restructure content, add schema and launch a link campaign in the same month, the resulting movement is unattributable. Sequence them and hold the observation window steady.
Add a citation layer. Track whether your brand is named inside generated answers, separately from whether anyone clicked. This is the measurement that the CTR decline makes necessary, and it is the one that does not exist in most Irish reporting today.
The Honest Limits of All of This
I want to be clear about what is not known, because the gaps matter as much as the findings.
Nobody has published a causal study of AI Overviews on named Irish publisher traffic. The figures circulating about publisher losses are American, British or global. Nobody has measured what share of Irish commercial search results is held by content originating outside Ireland, despite that claim being repeated constantly. And there is no independently audited Irish figure for the size of the SEO services market in euro, so the EUR 1.146 billion adspend number gets used as a proxy for something it does not measure.
Those are four holes in the evidence base for a single market. Any provider who fills them with confident percentages is guessing, and you should ask which primary source they are quoting.
What we do have is unusually good for a market of Ireland's size: one large first-party click study, one primary-source coverage measurement with a named methodology and date, and a clear, corroborated finding that the decline lands unevenly across query classes. That is enough to build a reporting framework on. It is not enough to build a forecast on, and the difference between those two things is where most of the overclaiming in this field happens.
The deeper argument, that visibility and citation are now separate disciplines requiring separate instruments, is the subject of Cited or Silent. If you want the framework without buying anything, the gated edition is free to download. Our own measurement work across markets is published in the AI Citation Rate Report 2026.
Written by Tessar Napitupulu, Founder and CEO of PT Arfadia Digital Indonesia, a member of the Forbes Agency Council, and author of Found Before They Search and Cited or Silent. Arfadia has documented its generative engine optimisation practice since 2023 and works from Jakarta, Bandung and Bali.
Frequently Asked Questions
Did AI Overviews cause the 26% drop in Irish click-through rate?
Not provably, no. The Friday.ie dataset that produced the 26% figure begins in February 2025, and AI Overviews reached Ireland in March 2025, so the study compares two periods that are both partly inside the new regime. The authors state this limitation themselves. The decline is real and measured on 175 million Irish searches. The attribution to a single cause is not established, and other factors including SERP layout changes, query mix shifts and competitor behaviour are not controlled for.
How common are AI Overviews in Ireland compared with the UK and the US?
Substantially less common. Ahrefs analysed 108 million AI Overview queries in November 2025 and found the feature on 12.00% of tracked Irish queries, ranking Ireland 44th of 50 countries. The United Kingdom sat at 19.10% and the United States at 20.50%. Ireland also ranked below Portugal, Denmark, the Czech Republic and New Zealand. Quoting an American figure to an Irish client overstates exposure by roughly 60 to 70 percent.
When did AI Overviews launch in Ireland?
Google announced the expansion on 25 March 2025 and rolled it out the following day, 26 March 2025. Ireland was one of nine additional European markets in that wave, alongside Austria, Belgium, Germany, Italy, Poland, Portugal, Spain and Switzerland. Irish users received the feature in English only, and at launch it was restricted to signed-in users aged 18 and over. France was excluded from that wave because of its stricter neighbouring-rights regime and remained without the feature into 2026.
Which types of query lost the most clicks?
Informational queries, by a wide margin. The Irish study found commercial-investigation terms holding a 4.24% click-through rate at positions four to ten, while informational terms at the same positions managed only 1.16%. That is a ratio of roughly 3.7 to 1. Ahrefs' separate trigger analysis corroborates it from the other direction, finding AI Overviews appear more often on informational, longer and higher-volume queries, and less often on branded, local and short ones.
Why do different studies disagree about zero-click search?
Because they define and measure it differently. A randomised field experiment on 1,065 desktop Chrome users in the United States found zero-click sessions rising from 54% to 72% between January and February 2026. Semrush, over an overlapping period, found zero-click share falling from 33.75% to 31.53%. Both cannot describe the same phenomenon. The phrase is used variously for sessions ending with no click at all, sessions where the click went to a non-organic element, and results pages with no organic link above the fold. Those definitions produce very different numbers from identical behaviour.
Is the 34.5% figure a zero-click statistic?
No, and this is a common misattribution. The 34.5% belongs to Ahrefs and measures click-through rate for the top organic result when an AI Overview is present, based on 300,000 keywords in April 2025. A December 2025 re-run widened it to roughly 58%. It is frequently and incorrectly credited to the Agarwal and Sen field experiment, whose zero-click finding is the move from 54% to 72%. If you see one study credited with both numbers, two studies have been merged.
Should we stop tracking rankings in Ireland?
No, but rankings should be reclassified. Position still determines whether you are eligible to be seen at all, so it remains a useful diagnostic. What the Irish data shows is that position no longer reliably predicts commercial outcome, because average position improved while click-through rate fell 26% over the same period. Treat ranking as an input metric reported in its own section, and keep click behaviour, downstream commercial results and citation presence in separate layers rather than blending them into one dashboard number.
Can we apply UK search benchmarks to an Irish account?
Not for this question. AI Overview coverage differs by seven percentage points between the two markets, so any target or forecast calibrated to British exposure is calibrated wrongly. Beyond the numbers, the institutional context differs: Irish content needs the Central Bank of Ireland rather than the FCA, the Data Protection Commission rather than the ICO, the Companies Registration Office rather than Companies House, Revenue rather than HMRC, euro pricing and Eircodes. Irish English follows British spelling conventions, so spelling alone signals nothing. The Republic and Northern Ireland should also be treated as two markets, not one.
What is happening to paid search spend in Ireland?
It is growing, but more slowly than everything around it. The IAB Ireland and IRM Online Adspend Study published on 14 May 2026 put total Irish digital adspend at EUR 1.146 billion for 2025, up 8%. Paid search accounted for EUR 345 million of that and grew 4%, against 12% for social and 14% for video. Search was the slowest-growing paid digital channel that year. One caveat worth knowing: Google, Meta, Twitter, Snapchat, Pinterest and LinkedIn do not submit data to the study, so their spend is estimated through econometric modelling rather than counted.
What has not been measured about Irish search?
Four things, and they matter. There is no published causal study of AI Overview impact on named Irish publisher traffic, so all publisher-loss figures in circulation are American, British or global. There is no measurement of what share of Irish commercial search results is held by content originating outside Ireland, despite the claim being widely repeated. There is no independently audited euro figure for the Irish SEO services market, which is why the adspend total gets misused as a proxy. And there is no Irish-specific benchmark for how long results take to appear. Any provider quoting precise numbers for these is estimating.
Sources & References:
- Friday.ie, Irish search click-through analysis, published 6 July 2026. 175 million Irish searches from Google Search Console over a 16 month window beginning February 2025. Aggregate CTR 3.62% to 2.67%, a fall of 26%; average position 6.38 to 6.18. Commercial-investigation terms 4.24% CTR at positions 4 to 10 against 1.16% for informational terms. The authors note the dataset begins after AI Overviews had already rolled out in Ireland, so no clean pre-feature baseline exists.
- Ahrefs Brand Radar, "Which Countries Have the Most AI Overviews", published 4 November 2025. 108 million AI Overview queries across 50 countries. Ireland 12.00% coverage from 1.4 million tracked keywords and 168,000 AI Overviews, ranked 44th; United Kingdom 19.10%; United States 20.50%; Indonesia 37.20% ranked first; Philippines 29.10% ranked second. English accounted for 52.75% of all AI Overviews in the dataset.
- Ahrefs, click-through rate analysis of 300,000 keywords, April 2025, finding a 34.5% CTR decline for the top organic result when an AI Overview appeared, with a December 2025 re-run widening the decline to approximately 58%. Also found AI Overviews appear more frequently on informational, longer and higher-volume queries and less frequently on branded, local and short queries.
- Saharsh Agarwal and Ananya Sen, randomised field experiment on 1,065 desktop Chrome users in the United States, January to February 2026, working paper revised 17 June 2026. Outbound clicks to publishers fell 39.8%; zero-click sessions rose from 54% to 72%. Desktop-only, United States only.
- Semrush, zero-click search measurement finding a decline from 33.75% to 31.53%, directionally contradicting other zero-click studies over overlapping periods. Included specifically to document that the evidence base is not settled.
- Google, "AI Overviews expand to more countries in Europe", announcement 25 March 2025, with rollout 26 March 2025 across nine additional European markets including Ireland, in English only and initially limited to signed-in users aged 18 and over. France was excluded from this wave.
- European Commission, preliminary findings that Alphabet may have breached the Digital Markets Act, issued approximately one week before the March 2025 AI Overviews expansion; formal antitrust investigation into Google's AI content practices opened 9 December 2025.
- IAB Ireland and IRM, Online Adspend Study 2025, published 14 May 2026. Total Irish digital adspend EUR 1.146 billion, up 8% year on year from EUR 1.06 billion. Paid search EUR 345 million, up 4%, representing 30% of the market; display EUR 735 million, up 10%; social EUR 576 million, up 12%; video 39% of total spend, up 14%. Study participants comprised 22 publishers and media owners, 2 ad networks and 7 media buying agencies. Google, Meta, Twitter, Snapchat, Pinterest and LinkedIn do not submit data; their spend is estimated through IRM econometric modelling.