Google's AI Overviews Are Cutting Website Clicks in 2026: What the Data Actually Shows and What to Do About It

Google's AI Overviews Are Cutting Website Clicks. Here's What the Data Actually Shows, and What to Do About It

If your website's traffic has fallen over the past year without anything on your end changing, Google's AI Overviews are a real and measurable part of the reason, not a theory someone in marketing invented to sell you something. Ahrefs tracked 300,000 keywords between December 2023 and December 2025 and found the top-ranking page gets 58 percent fewer clicks when an AI Overview sits above it. Pew Research Center, observing 900 people across nearly 69,000 real Google searches, found people click through 15 percent of the time normally, and 8 percent of the time when an AI-generated summary appears first. Two separate organisations, two different methods, the same direction, roughly the same size of effect. This is not a rumour. It is also not, as the rest of this piece will explain, the emergency most of the advice written about it makes it sound like.

This article covers three things properly, because most of what's been written on this topic in the past year covers none of them properly: how big the effect actually is and on what kind of search, what Google itself says you should do about it, which is quietly different from what most SEO blogs are telling you to do, and the practical, sequenced decisions worth making about a small business website in light of all this.

How Big Is the Actual Effect, and on What?

Start with what an AI Overview is. It's the block of text Google generates and places at the top of a search results page, summarising an answer before showing any of the ten blue links underneath it. It draws from Google's existing index, the same pages that would otherwise rank organically, and writes a synthesised answer using several of them at once, sometimes citing two or three, sometimes citing none visibly at all.

The Ahrefs and Pew figures above describe the average effect across all searches. The average understates how uneven the effect actually is, and this is the part almost every article on the subject glosses over in favour of a scarier headline number.

Informational searches are hit hardest

Queries that start with "what is," "how does," "why does," or "when should," the kind where a good two-paragraph answer genuinely resolves the person's question, are exactly the kind an AI Overview can fully satisfy without anyone needing to click anywhere. If your website has built traffic on general explainer content, blog posts answering broad questions in your industry, this is where you're feeling the drop, and it's a real, structural shift, not a temporary dip.

Commercial and transactional searches are affected far less

Queries like "book a plumber Newcastle," "quote for kitchen renovation," or a business's own name still send people through to a website, because an AI summary cannot take a booking, quote a job, or complete a purchase. It can describe the category of businesses that do those things. It cannot be the business that does them. If most of your traffic has always come from people actively looking to buy or enquire rather than people researching a general topic, the headline statistics above matter to you less than the coverage would suggest, and it is worth establishing which category you actually fall into before reacting to either the numbers or to this article.

Local and branded searches sit somewhere in between

Someone searching your business name, or "electrician" plus their suburb, is often close enough to a decision that an AI Overview either doesn't fully appear, or appears alongside Google's Business Profile and map results rather than replacing the click entirely. This is a separate mechanism from AI Overviews and worth understanding on its own terms rather than folding into this discussion, but it's the reason local businesses tend to describe this shift as less catastrophic than the raw CTR numbers imply.

What Google Itself Actually Says You Should Do, and Where the Popular Advice Gets It Wrong

Here is where most of what's currently written about this topic goes soft, and where the real gap sits. Search almost any article on "how to get cited in AI Overviews" published in the past year and you'll find close to the same three-step advice repeated: add FAQ schema markup, add HowTo schema markup, and keep your content structured for machines. It sounds authoritative. It is, at best, an oversimplification of what Google has actually published on the subject.

Google's own Search Central documentation says plainly that there is no special schema.org markup required for AI Overviews or AI Mode, and that structured data is not a requirement for appearing in generative AI search results at all. Google's guidance instead points back to the same fundamentals that have always mattered for ranking normally: helpful, accurate, well-organised content, clear crawlability, internal linking that helps both people and Google's systems understand how your pages relate to each other, and the same quality signals Google has described for years under its E-E-A-T framework, experience, expertise, authoritativeness, and trustworthiness.

This matters because a lot of small business owners are being sold, sometimes literally sold as a paid service, "AI Overview optimisation" that amounts to bolting FAQ schema onto a page and calling it done. Schema markup is not worthless. It genuinely helps machine-readable systems, including Google's own indexing and rich-result features, understand what's on a page, and there's no harm in using it correctly. But it is not the lever Google says actually determines whether your content gets pulled into an AI-generated answer. The lever is whether the underlying content is genuinely clear, accurate, and well-structured for a person to read, which a good writer and a well-organised page already achieve without any markup at all.

The honest translation of Google's own guidance, for a small business owner, is this: there is no shortcut here, no plugin, no line of code, that gets you cited. What gets you cited is writing an actual, direct, correct answer to the question your customer is asking, in your own words, on a page that's easy for both a person and a crawler to read. That was good practice before AI Overviews existed. It still is now. The one genuine addition worth making is answering questions in the same plain, direct phrasing a customer would actually search, rather than the marketing phrasing a business tends to default to.

Where This Leaves the Structure of the Website Itself

None of the above matters if the website physically can't be changed. This is the part the AI Overview conversation keeps skipping, and it is, in practice, the biggest single factor determining whether a small business can respond to any of this at all.

A meaningful share of small business websites in Australia were built once, handed over, and never touched again. There's no content management system running, or there is one nobody remembers the login for, and adding a new page, a new answer, or a new paragraph means going back to whoever built the site originally and paying by the hour, if that person can even be reached three years later. For a site like that, the advice to "add clear, direct answers to real customer questions" is sound advice aimed at a website that doesn't structurally exist yet. The website needs to become editable before it can become better.

This is the practical, unglamorous reason a properly built, actively maintained website matters more under this shift, not less. A full custom build through Dream Websites starts at $5,000, includes two months of support after launch, and runs on a structure where adding an answer, a page, or an update is a normal, ordinary task rather than a re-engagement with a contractor. The ongoing care that follows, $150 a month, covers updates, backups, and managed hosting, which in practice means the site is something a business can keep genuinely current, rather than something frozen at the moment it launched. In an environment where the difference between getting cited and not getting cited increasingly comes down to whether your content is current, clear, and correct, a website that's actually easy to keep current is not a luxury feature. It's the precondition for doing anything else on this list.

The Order to Work Through This In

First, work out whether your traffic loss, if you have one, is actually coming from informational searches or from something else entirely. Look at your analytics by query type if you have access, or think honestly about how customers find you. A drop concentrated in general explainer content with stable branded and commercial traffic points squarely at AI Overviews. A broad drop across everything, including branded searches for your own business name, points somewhere else, and shouldn't be blamed on AI Overviews by default just because it's the current explanation everyone reaches for.

Second, check whether your current website can actually be edited without a full rebuild. This is the honest starting point for almost everyone reading this, and it's worth being blunt with yourself about it. If adding a paragraph means finding a contractor and negotiating a scope of work, that is the structural problem to solve first, ahead of anything about schema markup or AI Overviews specifically.

Third, if the site can be edited, write clear, direct answers to the actual questions customers ask you, in the phrasing they'd actually use. Not keyword-stuffed copy written for a search engine that no longer parses pages that way, an honest, specific answer in plain language. This is the single change Google's own guidance points to most directly, and it's also, not coincidentally, the same thing that makes a page genuinely useful to a human visitor.

Fourth, use structured data where it's genuinely appropriate, not as a magic trigger. FAQ markup on a page that already contains real, plainly answered questions is sound practice. FAQ markup bolted onto vague marketing copy to try to game an AI system is not going to do what the advertisement for that service implies it will.

Fifth, put a genuine update cadence in place. A site that gets a real, useful change every month or two, a new answer, an updated price, a corrected detail, is in a materially different position over an eighteen-month horizon than one that hasn't been touched since the year it launched, regardless of what specific technique is used on any single page.

Sixth, don't rebuild everything this quarter because of a single trend. The fundamentals that have always made a business trustworthy to a person who lands on its website, a clear explanation of what you do, honest pricing information, an easy way to get in touch, real evidence of the work you've done, still carry the weight they always have. AI Overviews change how much traffic the top of the search funnel sends toward that page. They do not change what makes the page worth landing on once someone gets there.

An Honest Next Step

If you're not sure whether your website can be updated without a full rebuild, or which category of search traffic you're actually losing, if any, that's a short, genuinely useful conversation, not a sales pitch dressed up as one. Book a free strategy session and we'll look at your specific site and tell you plainly whether this is worth acting on now or something to simply keep an eye on.

FAQ

Is SEO dead because of AI Overviews?

No. What's changed is where the click goes for purely informational searches that an AI summary can fully answer on its own. Commercial and transactional searches, the ones that actually drive bookings and sales, still send people through to websites, because an AI summary cannot complete a transaction. Search visibility for those queries still works largely as it always has.

Do I need to add schema markup to my website to appear in AI Overviews?

Not according to Google's own published guidance, which states there's no special schema.org markup required for AI Overviews or AI Mode, and that structured data isn't a requirement for generative AI search results. Well-used FAQ or HowTo markup on genuinely clear content doesn't hurt, but it isn't the deciding factor, and treating it as one is where a lot of current advice overstates itself.

Should I rebuild my website just because of AI Overviews?

Only if you've established your current site genuinely can't be updated without one. If it can be edited, the sensible move is smaller, ongoing improvements to how clearly it answers real customer questions, not a full rebuild driven by a single search trend.

How do I know if AI Overviews are actually affecting my business, rather than something else?

Check whether the drop is concentrated in general, informational content while your branded and commercial search traffic has stayed roughly steady. If the decline is broad and includes people searching for your business by name, look for a different cause before assuming AI Overviews are the explanation.