AI Overviews changed what a Google search returns. If you spend money on Google Ads, the question is what that did to the paid click.
The cleanest evidence comes from a 2026 randomised field experiment by Saharsh Agarwal and Ananya Sen. When AI Overviews appeared, organic clicks fell 39.8 per cent, while clicks on sponsored results stayed flat. Real-world client data is less tidy. Seer Interactive found paid click-through rates also fell on AI Overview queries, but brands cited inside the AI answer did far better on paid.
Put together: paid has proved more resilient than organic, and the paid click increasingly depends on whether your business shows up in the answer.
Interest in the channel is rising, too. In Australia, “google ads” is searched 74,000 times per month, up 22 per cent year on year (DataForSEO AU, 30 August 2026).
This article walks through what the data shows, what it means for your paid budget, and how we at 21 Webs buy clicks in the AI era.
Key Takeaways
- In a 2026 randomised field experiment, organic clicks fell 38 to 39.8 per cent when AI Overviews appeared, while sponsored clicks stayed flat.
- Paid is not immune. Seer Interactive’s client data found paid click-through rate fell 68 per cent on AI Overview queries (organic: 61 per cent).
- In the same data, brands cited in the AI Overview saw a 91 per cent higher paid click-through rate. Ads now appear above, below and inside AI Overviews.
- “Google ads” runs at 74,000 monthly searches in Australia, up 22 per cent year on year. “Google ads for small business” rose 467 per cent.
- Google is pushing automation through Performance Max and AI Max. It works best chosen campaign by campaign, with a human setting strategy.
- In SearchLight Digital’s US benchmark, Local Services Ads leads cost $53 USD on average, against $104 USD for standard Search Ads.
- Google says AI Mode passed one billion monthly users a year after launch, and it is testing new ad formats inside AI Mode.
Paid Clicks Held Up Better Than Organic When AI Overviews Arrived
The strongest evidence so far comes from a controlled experiment, and it favours paid.
Researchers Saharsh Agarwal (Indian School of Business) and Ananya Sen (Carnegie Mellon University) ran a randomised field experiment with 1,065 US desktop Chrome users between January and February 2026. They used a custom browser extension to randomly show or hide AI Overviews during live Google searches, isolating the causal effect for the first time.
The results, published on SSRN in April 2026 and revised June 2026, were stark:
- Organic clicks fell 38 to 39.8 per cent on queries where AI Overviews appeared.
- Zero-click searches rose from 54 to 72 per cent. Most users never left the results page.
- Clicks on sponsored (paid) results stayed flat. In this experiment, the loss sat in organic.
- User satisfaction scores did not change. Removing AI Overviews made no measurable difference to how users rated their search experience.
That last point is worth pausing on. Users did not miss the AI summaries when they were taken away, yet those summaries took nearly 40 per cent of organic clicks when present.
That tells you where this is heading. Google has no reason to remove a feature users rate the same either way and that keeps people on the results page. Plan on the assumption that AI Overviews stay and expand.
The zero-click shift matters as much as the click decline. When zero-click searches move from 54 to 72 per cent, the audience has not disappeared. It has stopped travelling. For an organic listing that is a serious problem, because the value of a ranking was the visit. A sponsored listing earns its keep differently: the impression, and the click that comes when the user wants a quote rather than an explanation.
For businesses that rely on organic search, this is a structural loss. For businesses running paid campaigns, the picture is better, though not untouched.
Other studies do not all point the same way. Ahrefs analysed 300,000 keywords using Google Search Console data and found AI Overviews correlate with a 58 per cent lower click-through rate for the top-ranking page. Seer Interactive tracked 3,119 informational queries across 42 organisations from June 2024 to September 2025 (25.1 million organic and 1.1 million paid impressions). On AI Overview queries, organic click-through rate fell 61 per cent and paid fell 68 per cent.
So paid is not immune. The experiment isolates one effect over a few weeks; Seer follows real accounts over fifteen months while the whole results page changed. The fair reading: paid held up better than organic under controlled conditions, and both are under pressure in the wild.
If you have been wondering whether the PPC versus SEO calculus changed in 2026, it did. Paid search is no longer just the “quick wins” channel. On current evidence it is the more resilient of the two, though not a guaranteed one.
The Channel Is Being Re-Bought, Not Abandoned
The search demand data tells its own story.
In Australia, “google ads” runs at 74,000 searches per month with a 22 per cent year-on-year trend rise (DataForSEO AU, 30 August 2026). People do not research a channel this heavily when they are walking away from it. Our reading is that many are researching it before they commit budget again.
The more revealing signal sits in the long tail. “Google ads for small business” rose 467 per cent year on year in the same dataset. That is a large jump, and it points to small business owners and marketing managers actively looking for how to use the platform.
Here is what we think is driving the demand.
Organic uncertainty. Business owners who built their lead pipeline on organic traffic saw it erode, sometimes without warning, as AI Overviews expanded. Paid search offers a floor: you pay, you appear.
AI-driven simplification. Google has made campaign setup easier with Performance Max and AI Max. The barrier to entry dropped while the urgency rose.
Visibility anxiety. When your competitor appears in the sponsored slot above the AI Overview and you do not, the gap is visible and immediate.
For a Werribee-based plumber or a Melbourne accountant, the question is not “should I be on Google Ads.” The question is “how much of my budget should shift there now that organic is less predictable.”
There is a useful way to think about that. Split your lead sources into two piles. Pile one is everything you own and earn: organic rankings, your Google Business Profile, referrals, repeat customers. Pile two is everything you rent: paid clicks, marketplace leads, anything where the flow stops the moment you stop paying. The AI shift hit pile one harder than pile two. Do not abandon pile one; owned channels are still your cheapest leads. Just stop treating it as enough on its own.
We wrote about the broader zero-click economy and Australian organic search. Paid is largely the response to that shift.
Ads Now Live Inside the AI Answer, and Cited Brands Do Better on Paid
Google is now selling ad space inside the AI answer itself. That changes where a paid click comes from.
According to Google’s own Ads Help documentation, ads are now eligible to appear above, below and within AI Overviews. Australia is one of the markets where ads inside AI Overviews are already live, in English, on mobile and desktop.
Here is what that means in practice:
- Text and Shopping ads from existing Search, Shopping and Performance Max campaigns can appear within AI Overviews without any special opt-in or new campaign type.
- Google uses both the user query and the content of the AI Overview to decide which ads to serve.
- Sensitive verticals (Google lists adult, alcohol, gambling, finance, healthcare and politics, among others) are excluded from ads inside AI Overviews.
Seer Interactive’s data adds a layer most advertisers have not absorbed yet. In the same study that found paid click-through rate falling 68 per cent on AI Overview queries, brands cited inside the AI Overview earned a 35 per cent higher organic click-through rate and a 91 per cent higher paid click-through rate than brands that were not. It is a correlation in client data, not a controlled test, but it suggests AI citation and paid performance now move together.
In plain terms: a cited brand saw close to double the paid click-through rate. Citation may not be the only cause, since cited brands tend to be better known, but it is the clearest lever you can work on.
The person reading an AI Overview is informed, not converted. What they need next is a business they can trust. The citation does the persuading. The ad collects the intent.
This is why we tell our clients at 21 Webs that AI visibility and paid search are not separate strategies. They are the same strategy. Your content earns the citation. Your ad captures the click. The two work as a pair.
So whoever runs your ads and whoever writes your content should be looking at the same queries. A practical check: take the search terms that trigger AI Overviews in your category and ask two questions. Is the brand mentioned in the answer? Is there an ad in the placement? A mention without an ad is an open door. An ad without a mention is paying full price for a harder conversation.
We covered the mechanics of how AI search reshapes local discovery in a recent piece. The paid implications sit right alongside the organic ones.
Performance Max and AI Max Made the Platform an AI-Bidding Machine
The Google Ads account most businesses set up five or six years ago looks very different today.
Two developments reshaped what an advertiser actually controls:
Performance Max runs across Search, YouTube, Display, Gmail, Maps and Discover from one campaign. Google’s system decides placement, creative and bids in real time. Channel-level reporting now shows more of where the budget went, but the allocation stays with the system.
AI Max for Search left beta in April 2026, and Google says hundreds of thousands of advertisers already use it. Campaigns using automatically created assets or campaign-level broad match are being upgraded to AI Max from September 2026; the Dynamic Search Ads upgrade was pushed back to February 2027. AI Max brings three features:
- Search term matching: the system infers intent from your landing pages and ad copy, matching your ads to queries you never explicitly bid on.
- Text customisation: generative AI writes headline and description variations in real time to match each query.
- Final URL expansion: Google picks the landing page it predicts will suit each search, even if you did not nominate it.
Google says AI Max for Search campaigns see an average of 7 per cent more conversions or conversion value at a similar cost per acquisition or return on ad spend when using the full feature suite. That is Google’s own figure; your result will depend on your data and landing pages.
What does this mean for the advertiser’s job?
- Keywords become one signal among several. Landing pages, audience lists, conversion goals, brand guidelines and negative keywords all steer the system.
- Ad copy becomes guardrails as much as writing. Pinned text, brand voice and excluded terms decide what the system is allowed to generate.
- Outcomes matter more than inputs. The metric that matters is cost per qualified lead, not individual keyword bids.
Parts of this are good news. Managing 300 keywords with match types and bid adjustments never suited a business owner’s time, and automation can take on much of it. But it is a tool, not a default. Whether a campaign suits Performance Max, AI Max or a tighter Search setup with conversion-focused bidding depends on its data, budget and goals, and the answer can change. No automated system can define your offer, your pricing or the landing page that converts. Those stay human decisions.
Two of those three features deserve a closer look, because they are where accounts quietly go wrong.
Final URL expansion is the one that catches people out. The system chooses a landing page for each query, and whichever page it picks is where your money lands. On a well-structured site that helps. On a site that is a maze, it is a liability, because the system is making navigation decisions with your budget. Switching it off is rarely the fix. Making every page it might pick able to convert on its own usually is.
Search term matching does something similar to your targeting, matching you to queries you never bid on. That widens your reach and your exposure together, which turns the negative keyword list from housekeeping into strategy. If your page talks about emergency call-outs, the system will find emergency queries. Decide in advance which of those you are happy to pay for.
If your account was built years ago around exact-match keywords and manual CPC bidding, it is worth a fresh look, whether the answer turns out to be more automation or less. Talk to our team about what suits your campaigns.
"Google Ads for Small Business" Is Up 467 Per Cent Year on Year
The search data shows a sharp rise in small business interest.
When “google ads for small business” grows 467 per cent year on year (DataForSEO AU, 30 August 2026), it suggests owners who sat on the sideline are now looking for a way in.
We see a few likely reasons:
The organic traffic squeeze. Small businesses that relied on local SEO and blog content for leads are seeing declining click volumes. Paid search becomes the backstop.
AI-simplified onboarding. Google’s guided campaign setup, combined with Performance Max and AI Max, means a business owner can launch a campaign without understanding match types, quality score mechanics or bid modifiers.
Competitive pressure. When a competitor’s ad sits above the AI Overview and yours does not, they are first in line for the enquiry. In a crowded local category, that gap is hard to ignore.
For an Australian small business, the practical question is budget, and there is no universal right number. It depends on your category’s click costs, your service area and how well your landing page converts. A useful test: does the budget buy enough clicks each month to learn from?
The risk is running the channel poorly: broad targeting with no negative keywords, traffic sent to a homepage instead of a purpose-built landing page, or Google’s “smart” defaults left unreviewed.
None of those three failures is fixed by spending more. We have written about what AI search means for SEO; the paid side needs the same strategic hand.
Local Services Ads Can Deliver Leads at Roughly Half the Cost of Search Ads
For Australian service businesses, Local Services Ads (LSAs) add another rung to the paid ladder.
You pay per lead, not per click: a call, message or booking through the ad. If a lead is junk (wrong area, wrong service, repeat caller), you can dispute it and may receive a credit. The ad sits at the very top of the results page, above standard text ads and the Map Pack, with a Google Verified badge.
The most detailed public benchmark we have found is from the US, so read these as US dollars and markets:
Local Services Ads: SearchLight Digital’s February 2026 US home-services benchmark covered 888 contractors, $6.72 million USD in LSA spend and 126,650 leads. The blended LSA cost per lead was $53 USD. By trade, electrical came in at $39, HVAC at $51, plumbing at $57 and drain and sewer at $59 (all USD).
Standard Google Search Ads: In SearchLight’s January 2026 Google Ads sample, standard Search Ads averaged a blended $104 USD per lead, with a wide split: branded search ran at $34, non-branded search at $149 and Performance Max at $72 (all USD).
On those numbers, the blended LSA lead cost about half the blended Search lead, and the gap is wider against non-branded search, where most new-customer demand sits. Branded search is cheaper than both, because that searcher already knows you. Australian costs will differ by city, category and competition, and we have not seen a published Australian dataset at this scale. Treat the US figures as a guide to the gap, not a local price list.
The practical play for most Australian service businesses is to run both formats:
LSAs capture high-intent callers, usually at a lower cost per lead. They work especially well for emergency and “near me” searches where the user wants to call, not research.
Search Ads capture broader demand. They let you target specific keywords, run dedicated landing pages, and reach services that LSAs may not cover in your category.
Think of it as a ladder rather than a choice. The top rung is the emergency caller who needs someone now, and LSAs are built for that person. The rungs below are researched purchases with a decision window measured in weeks, and those belong to Search Ads and your own pages. Run only one rung and you either overpay for emergency leads or miss them.
In Australia, LSA eligibility depends on your category and region, and not every service type is available yet. Screening (licence, insurance and background checks, depending on the category) can take several weeks. If you qualify, start early: review volume and response time feed into placement.
Plan around that screening window rather than react to it. An early start also compounds: more placement brings more leads, more leads bring more reviews, and more reviews lift placement again.
If you are a Werribee plumber or a Melbourne electrician and you are not yet running LSAs alongside your Google Ads campaigns, you may be missing some of the lowest-cost leads available in your category.
How 21 Webs Buys Clicks in the AI Era
We run our own paid campaigns. We eat what we cook.
Here is how we approach paid search at 21 Webs today:
We use automation selectively, and a human sets the strategy. Bidding strategy varies by campaign and changes over time, but we rely mainly on conversion-focused bidding such as Maximize Conversions, with hands-on management around it: clear conversion goals, tight negative keyword lists and landing pages that match the search. Performance Max and AI Max are tools we assess on their merits, not defaults we hand an account to. The bidding strategy stays a human decision, judged campaign by campaign.
We pair paid with AI-citation strategy. The data suggests ads work harder when the brand also appears in the AI Overview. So the content that earns citations, through our SEO work, and the ads that capture the click are planned together by one team. Each client has a human lead who owns those decisions, backed by AI agents doing research and monitoring.
We watch cost per qualified lead, not cost per click. A cheap click that bounces is worth nothing. A click that costs several times more and turns into a quote request can be worth far more. Our reporting is built around business outcomes, not platform vanity metrics, and clients get it every week.
We invest in landing pages as much as ad spend. AI Max can generate headlines and select URLs, but it cannot fix a poorly structured page. We build our landing pages in-house, with fast load times, mobile-first layouts and clear calls to action, to the same standard we build for clients through our web design service.
We set up Local Services Ads for eligible clients. Where the category qualifies, the cost per lead can sit well below standard Search Ads, as the US benchmark above suggests. We treat LSAs as one rung of the paid ladder, with Search campaigns picking up the queries LSAs cannot cover.
What we do not do is set and forget. Any automated campaign needs a human reading the search term reports and flagging spend on terms that do not convert. Good practice also means checking that spend by time of day and location still matches where the leads come from.
Without that cadence, an AI-run account drifts away from the business it was meant to serve. The drift is invisible in a dashboard and obvious in a bank account.
The AI runs the auction. We run the strategy.
Chart: Organic vs Paid Click Impact When AI Overviews Appear
Source: Agarwal and Sen (2026), randomised field experiment, 1,065 participants, January to February 2026. Published on SSRN, revised June 2026.
This chart shows the controlled experiment only. Seer Interactive’s observational client data, covered above, found paid click-through rate also fell on AI Overview queries.
Metric | Without AI Overview | With AI Overview | Change |
|---|---|---|---|
Organic clicks per search | 0.61 | 0.38 | Down 38 to 39.8% |
Paid (sponsored) clicks per search | Flat baseline | Flat baseline | No significant change |
Zero-click rate | 54% | 72% | Up 18 points |
Important FAQs
Do Google Ads still work?
Did AI Overviews kill paid search?
What happened to Google Ads?
Is Google Ads worth it in the AI era?
The Bottom Line
Google Ads is not a search channel anymore. It is an AI-placement channel that happens to live on a search engine.
Paid clicks held up better than organic through the change that hit organic hardest. Interest in the channel is rising in Australia, with 74,000 monthly searches for “google ads”, up 22 per cent. AI Overviews created new ad placements, and in Seer Interactive’s data, brands cited in the AI answer saw a 91 per cent higher paid click-through rate.
Picking keywords, writing three headlines and setting a daily budget no longer covers the job. The work has moved towards strategy, signal quality and landing pages, whatever campaign type you run.
For Australian small businesses, four moves make sense:
Run paid alongside SEO. Paid is your most dependable floor. SEO earns the AI citations that help your ads work harder.
Check Local Services Ads. If your category qualifies, the cost per lead can sit well below standard Search Ads.
Invest in landing pages. The AI can bid and write headlines. It cannot fix a page that does not convert.
Watch outcomes, not inputs. Cost per qualified lead is the number to manage to.
Is your Google Ads account built for the AI era?
We are a Werribee-based Google Partner agency, and we run paid search for ourselves and Australian clients every day. Book a free Google Ads audit and we will show you what to keep, what to fix and where automation is earning its place, or see how our Google Ads management works.
Sources
[1] Agarwal, S. (Indian School of Business) and Sen, A. (Carnegie Mellon University), “The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment”. Randomised field experiment, 1,065 US desktop Chrome users, January to February 2026. Published on SSRN, April 2026; revised June 2026.
[2] Seer Interactive, “AIO impact on Google CTR”, November 2025. 3,119 informational queries, 42 organisations, June 2024 to September 2025; 25.1 million organic and 1.1 million paid impressions.
[3] Ahrefs, study of AI Overviews and click-through rate, February 2026. 300,000 keywords, Google Search Console data.
[4] SearchLight Digital, US home-services lead cost benchmark, February 2026. 888 contractors, $6.72 million USD Local Services Ads spend, 126,650 leads.
[5] Google, “Ads in AI Overviews”, Google Ads Help. https://support.google.com/google-ads/answer/16297775?hl=en
[6] Google, “We’re upgrading Dynamic Search Ads to AI Max”, 15 April 2026, updated 11 June 2026. https://blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026/
[7] Google, “A new era for AI Search”, Elizabeth Reid, VP Search, 19 May 2026. https://blog.google/products-and-platforms/products/search/search-io-2026/
[8] Google, “A new generation of ads for the AI era of Search”, 20 May 2026. https://blog.google/products/ads-commerce/google-marketing-live-search-ads/
[9] DataForSEO, Australian keyword data for “google ads” and “google ads for small business”, pulled 30 August 2026 (21 Webs’ own data pull).
