How AI Overviews Are Changing Paid Search CPCs

AI Overviews have not made search ads uniformly cheaper or dearer. They have redistributed where clicks happen and what each one is worth. Google's own filings show paid clicks up 13% and CPC up 3% — while CTR on informational AI Overview queries collapsed. Here is what is actually happening, and how to bid for it.

AI Overviews have not made Google search ads cheaper or dearer across the board. They have redistributed where clicks happen and what each click is worth. Google's own filings show paid clicks up 13% and cost-per-click up just 3% year on year in Q2 2026. At the same time, click-through rates on informational queries with an AI Overview collapsed through 2025, then partly recovered in early 2026. Both things are true, and the gap between them is where advertisers are currently losing money.

This guide works through the real data — Google's SEC filings, Seer Interactive's two large studies, Adthena's industry analysis — and what it means for how you bid, which keywords you keep, and where budget should go next.

The numbers at a glance

What Google's own filings say

Start with the most authoritative source, because it contradicts the loudest claims. If AI Overviews were destroying paid search, it would show in Alphabet's monetisation metrics first. It does not.

Paid clicks grew double digits, a full year into AI Overviews and after AI Mode's wider rollout. Cost-per-click rose, but modestly — nothing like the "CPC inflation crisis" some commentary describes. Google's CEO has repeatedly said AI features are driving search query growth, and the query growth is the part that explains rising paid clicks.

Two caveats the headline hides

First, growth is decelerating. Search revenue growth slowed from 19% in Q1 to 17% in Q2 — the first step down in six quarters. That is not a collapse, but it is a signal worth watching rather than dismissing.

Second, revenue holding up is not the same as your clicks holding up. Google's figures are aggregate. They blend every advertiser, every query type, every device and every country. A 3% average CPC increase can hide a 25% increase on the bottom-funnel keywords your campaigns actually depend on, offset by falling prices on informational terms you were never profitable on anyway.

The Seer story: a collapse, then a reversal

Seer Interactive has published the most-cited research on AI Overviews and paid search, and the most useful thing about it is that they revised their own conclusions publicly.

The 2025 study: a 68% drop

Seer's September 2025 analysis covered 3,119 informational and educational search terms across 42 client organisations — 25.1 million organic impressions and 1.1 million paid impressions from June 2024 to September 2025. On queries showing an AI Overview, paid CTR fell from 19.70% to 6.34%. Organic CTR on the same queries fell 61%.

That finding spread quickly and set the narrative most marketing teams are still budgeting against. Seer's own analyst said she was not optimistic about a rebound.

The 2026 update: the prediction was wrong

Five months later, with a much larger dataset — 5.47 million tracked queries and 2.43 billion organic impressions, spanning informational, transactional and commercial intent — Seer published an update that marked their own prediction as wrong.

Read those two lines together, because the combination is the point. SERPs with an AI summary got better for paid advertisers over the period. SERPs without one got worse. Paid clicks appear to be migrating toward AI Overview results rather than disappearing from them.

Organic followed a similar pattern: CTR on AI Overview queries climbed from a floor of 1.3% in December 2025 to 2.4% in February 2026.

What the reversal does and does not mean

It does not mean AI Overviews stopped suppressing clicks. AIO-present paid CTR is still well below non-AIO paid CTR in the same dataset — 16.21% against 21.85%. The gap narrowed; it did not close.

It also does not mean the 2025 finding was false. It applied to informational queries specifically, where paid ads have always struggled. The 2026 dataset is broader, and includes the commercial and transactional queries where AI Overviews compete less directly with ads.

What it does mean is that anyone still planning 2026 budgets on a "68% collapse" assumption is working from a stale number. The honest current picture is compression, not collapse — and compression that varies sharply by query type.

Why CTR can fall while CPC barely moves

This is the part that confuses most teams. If fewer people click ads on AI Overview queries, why is average CPC only up 3%? Four mechanisms explain it.

1. Budget concentrates on bottom-funnel queries

When informational queries stop converting, advertisers pull budget from them and put it on commercial and transactional terms. Informational CPCs soften as demand leaves; bottom-funnel CPCs rise as demand arrives. The average barely moves while both ends shift substantially.

2. Query volume is growing

More total searches means more total ad opportunities. Google has said AI features are driving query growth. Paid clicks can grow even while CTR on any given query falls, provided query volume rises faster.

3. The auction clears on fewer, more motivated clicks

Someone who reads an AI Overview and still clicks an ad has usually already decided they want to act. Those clicks convert better, so advertisers can afford to pay more for them. Higher conversion rates support higher CPCs without breaking return on ad spend.

4. Mix effects hide everything

Google's CPC figure is affected by device mix, geographic mix and currency. A shift in traffic toward cheaper markets or mobile can pull the average down while prices in your market rise. Google itself lists device mix, geographic mix and exchange rates among the drivers of CPC change in its filings.

Directional summary synthesised from Seer, Adthena and Google filings. Your account will differ by vertical and market.

How AI Overviews change the page

An AI Overview is a generated summary at the top of results. Where your ad sits relative to it now matters more than average position ever did.

The practical consequence: a long AI Overview can push a "position 2" ad below the fold on mobile. Your average position report says nothing changed. Your impressions say the same. Your CTR tells the truth.

Google is also moving deliberately into comparison and instructional territory. Adthena's analysis across six industries — covering more than 5 million ads between late December 2025 and January 2026 — found that in telecom, technology and retail, AI Overviews are dominated by comparison content. When Google presents a side-by-side comparison itself, the research phase is satisfied on the results page, and a click to your comparison landing page becomes less necessary.

Healthcare and finance see mostly informational Overviews, which act as a buffer — they answer the question without directly competing with the commercial next step. That is a meaningful difference for advertisers in those sectors, where the Overview educates and the ad still captures.

Short queries are not safe

A common assumption is that short, broad queries escape AI Overviews. The data does not support it. Seer found single-word informational queries still trigger an Overview 27.3% of the time. And Adthena reports that more than 60% of ad appearances inside AI Overview environments happen on three-to-four-word queries — ordinary mid-tail searches, not long conversational prompts.

Ads inside AI Overviews and AI Mode

The biggest structural change is not the Overview itself but Google placing ads within AI-generated answers.

Which campaigns are eligible

You do not need a special campaign type. Eligible campaigns for AI Overview and AI Mode placements include:

• Performance Max
• AI Max for Search, with search term matching
• Shopping campaigns
• Broad match Search campaigns

The AI Mode test has so far been limited to US advertisers on English-language queries, and early phases excluded ads using pinned assets or text disclaimers. That exclusion matters for regulated advertisers — finance, insurance, legal and healthcare often depend on disclaimers, and may find themselves ineligible for these placements even when they want them.

The new formats

Google Marketing Live 2026 framed AI Mode advertising around several new placements, including conversational discovery ads, highlighted answers, and an in-ad AI agent. It also introduced agent tooling across Google's ad products and a persistent cross-surface shopping cart spanning Search, the Gemini app, YouTube and Gmail.

The direction is clear even though the performance data is not yet in: the search ad is becoming part of the AI answer, and what Gemini says about your business in a paid placement increasingly depends on what your site, feed and assets say.

The citation effect

One of the most useful findings in the research is that being cited inside an AI Overview appears to lift paid performance on the same query.

Treat this carefully. Seer are explicit that it shows correlation, not causation. Brands with more authority are more likely to be cited, so the premium may reflect brand strength rather than the citation itself — and Seer cannot separate the two with their dataset.

Even with that caveat, it points somewhere practical. Organic visibility in AI answers and paid performance are no longer separate problems owned by separate teams. The citation sits upstream of the click. If your SEO team and your paid team do not talk, you are managing one outcome with two disconnected dashboards.

The measurement problem

This is the most frustrating part for anyone managing spend, and it is worth being blunt about it.

Google currently gives no placement-level reporting for AI Overviews or AI Mode. There is no segment, no filter and no impression-share breakout telling you which impressions came from an AI surface. You cannot see directly whether your ad appeared inside an Overview, beside it or below it.

The workable approach is to tag your top queries by whether they currently trigger an AI Overview, using a SERP tracking tool, and then compare CTR, CPC and conversion rate between the two groups in your own reporting. It is manual, but it is the only way to see the effect in your own account rather than in someone else's study.

Signals that AI Overviews are hitting you

• Falling CTR with stable or rising impressions. You are still being served; people are just not clicking.
• Rising CPC without a conversion-rate improvement. You are paying more for clicks that are not better.
• Shrinking top impression share on informational terms.
• Weaker engagement on informational landing pages tied to paid campaigns.
• Unexplained performance change in broad match or Performance Max after an AI Overview expansion.

A warning about AI Max

Because AI Max for Search is a main route into AI Overview and AI Mode placements, many advertisers are switching it on. Independent testing suggests caution.

Across 23 AI Max tests, accounts using all three AI Max features together outperformed those using single features, and text customisation lifted relevance and Quality Score in most tests. But 84% of advertisers saw neutral or negative AI Max outcomes when measured at account level, and gains were smaller at account level than at campaign level. Much of the apparently new traffic had been cannibalised from existing campaigns rather than representing new demand.

CPC by industry in 2026

Aggregate CPC barely moved, but industry-level CPC moved a great deal — in both directions. The 2026 benchmark data across 23 industries puts average CPC at $5.42, CTR at 6.64% and conversion rate at 8.18%, with sharply different movements underneath.

The education figure is telling. Education is an informational-heavy category — people researching courses, qualifications and subjects — exactly the kind of query AI Overviews answer well. As informational clicks lost value, demand and prices fell. Real estate, by contrast, is transaction-heavy and locally specific, where AI Overviews are less able to satisfy the need and advertisers keep bidding up.

That pattern is the most useful thing to take from industry data: the more of your category's searches are informational, the more AI Overviews reshape your auction.

What this means for Indian advertisers

Most published AI Overview research is US-based, and it is worth being clear about how far it transfers.

Three practical implications. First, the AI Mode ad formats that US advertisers are testing are not yet broadly available in India, so do not rebuild campaign structure around them. Second, the underlying mechanism — informational clicks losing value, budget concentrating on bottom-funnel terms — applies in any market where AI Overviews appear. Third, because absolute CPCs are far lower in India, the cost of testing is lower too. That makes India a reasonable place to measure your own AIO-versus-non-AIO performance before applying lessons to more expensive markets.

The knock-on effect on publishers

AI Overviews do not only affect advertisers. They affect the websites that ads appear on, and that side of the story is showing up in Google's numbers too.

In Q2 2026, Google Network revenue fell year on year, which Alphabet attributed primarily to a decrease in AdSense revenue. Network impressions fell 12% while cost-per-impression rose 13%.

The likely connection: when AI Overviews answer questions on the results page, fewer people click through to publisher sites, and publishers serve fewer ad impressions. For anyone who monetises content with AdSense, this is the same shift seen from the other side. It is a strong argument for publishers to reduce dependence on informational search traffic alone.

What to do: a working strategy

1. Audit your keywords by intent and AIO status

Pull your top 200 keywords by spend. For each, record two things: its intent (informational, comparison, commercial, transactional, branded) and whether it currently triggers an AI Overview. This single table will tell you more about your exposure than any study.

2. Segment campaigns by intent

If informational, commercial and transactional keywords share a campaign, they share a budget and a bidding strategy. That means the algorithm cannot price them differently, and your spend drifts toward whatever converts cheapest on paper. Separate them so each can be bid on its own economics.

3. Run the CPA maths honestly on high-funnel terms

Informational paid clicks may have partly recovered, but they recovered at a persistently low level. A stable CPC on a keyword with a collapsed conversion rate is a more expensive keyword, even though nothing in the CPC column changed. Your CPC has not moved; your efficiency has.

4. Bid for placement above the Overview on commercial terms

On the terms that matter commercially, the position above the AI summary is worth considerably more than a position below it. Average position alone no longer captures that — watch top and absolute-top impression share instead.

5. Make landing pages worth leaving the Overview for

Someone who clicks past an AI summary has already read the basics. A landing page that repeats what the Overview said gives them no reason to stay. Lead with what the summary cannot provide: pricing, availability, a calculator, a specific offer, proof.

6. Work on citation, not just bids

Given the correlation between being cited and higher paid CTR, organic content that earns citations is now part of paid performance. Clear answers, verifiable facts, structured data and genuine authority on your topic are what AI systems cite.

7. Change what you measure

Raw clicks are becoming a weaker success measure on AI-affected queries. Track share of voice, branded search growth, assisted conversions and AI Overview citations alongside click metrics — and judge every change at account level, not campaign level.

Where budget goes if search tightens

If AI Overviews make the top of the search funnel less efficient, the demand does not vanish — it just stops being cheaply capturable through search. Advertisers are responding in three ways.

Concentrating search on the bottom of the funnel, where intent is strongest and AI Overviews compete least, and accepting higher CPCs there because conversion rates justify them.

Rebuilding top-of-funnel demand elsewhere — creator video, short-form social, programmatic audio and native advertising, where awareness is generated rather than captured. These channels create the branded searches that later convert efficiently on Google.

Treating AI visibility as its own channel, with citation tracking, AI-answer optimisation and, where available, paid placement inside AI answers.

For advertisers whose informational search campaigns are losing efficiency, native advertising is one practical alternative worth testing for top-of-funnel reach. Native reaches people while they are reading content rather than while they are searching, which makes it largely independent of what AI Overviews do to the results page. Our native advertising CPC benchmarks show what that traffic costs, and the comparison calculator models it against your current numbers.

What happens next

Much of this change is being driven by a small group of companies — our list of the hottest AI startups in Silicon Valley covers who they are.

Three developments look likely based on current direction.

AI search ad spend grows fast from a small base. eMarketer forecasts US AI search ad spend rising from $2.08 billion in 2026 to $25.93 billion by 2029 — from 1.3% to 13.6% of total search ad spending. That is rapid growth, but it also means classic search still carries the overwhelming majority of spend for years yet.

AI Overview coverage expands into more commercial and mid-funnel queries. Expect exposure to depend less on static rank and more on how well campaigns align with the intent the AI interprets — which puts more weight on feed quality, structured data and asset depth.

Reporting improves, slowly. The absence of placement-level data is the most common complaint from advertisers. Google has strong commercial reasons to fix it eventually, but there is no announced timeline, so build your own measurement now rather than waiting.

The broad conclusion is less dramatic than most coverage suggests, and more useful. AI Overviews have not broken paid search. They have made its economics less uniform — rewarding advertisers who price each query on its own merits, and quietly taxing those still bidding on every keyword as if the results page looked the way it did in 2023.

Figures come from Alphabet's SEC filings and earnings releases, Seer Interactive, Adthena, WordStream, BrightEdge, Ahrefs and eMarketer, current to September 2026. Seer's 2025 and 2026 studies use different datasets and are not a single trend line. Benchmarks are directional — validate against your own account.