If you've searched anything on Google recently, you've probably noticed it. The clean list of blue links is no longer the default experience — instead, you're met with an AI-generated summary, a conversational follow-up box, or an invitation to "dive deeper" into AI Mode. This isn't a small tweak. Google AI Search has been labelled the biggest change to Search in over 25 years, and at Bidmark, we've spent the last several months watching it reshape client traffic, rankings, and strategy in real time.
We're not writing this as outside observers. We're in client accounts daily — restructuring content, rebuilding schema, auditing crawl health, and tracking how AI Overviews and AI Mode pull from (or ignore) the sites we manage. This is a longer read than our usual post, because "google ai search" isn't one feature — it's an umbrella covering several distinct products, each with its own mechanics and its own implications for anyone relying on organic traffic. We wanted to actually explain how it works, not just flag that it exists.
A quick bit of context: how we got here
Google has been layering AI into Search gradually for a couple of years, but what's happening now is different in scale. The company has rebuilt the core search bar around a newer Gemini model, moving away from the idea of a query returning a static results page and toward the idea of a query starting a conversation. Industry press covered the announcement bluntly — one widely shared tech headline described it as "Search as you know it is over" — and that's not hyperbole so much as a fair summary of the intent behind the change.
For context on scale: Google has said AI Overviews now reach more than 2.5 billion monthly users, and the standalone Gemini app has grown to around 900 million monthly active users. Gemini's share of overall generative AI traffic has also grown sharply over the past year, according to third-party traffic analysis. This isn't a beta feature tucked away in a menu anymore — it's the default way thousands of businesses' potential customers now search.
AI Overviews vs. AI Mode — they are genuinely different products
A lot of confusion we see from clients starts here, so it's worth being precise.
AI Overviews appear at the top of a normal results page, sitting above (or interspersed with) the traditional blue links. Google's systems decide, query by query, whether an AI-generated summary is likely to help — typically for broader or exploratory questions rather than narrow, transactional ones. Overviews include links out to supporting sources, and Google is explicit that the responses can contain mistakes, which is why it encourages people to check more than one source and to use the thumbs up/down feedback on each summary. You can filter a specific search to show only web links using the "Web" filter, but AI Overviews can't be switched off site-wide — they're treated as a core Search feature, the same way knowledge panels are.
AI Mode is a much fuller experience, closer in feel to a chatbot than a search engine. Instead of returning one static answer, it breaks a question into subtopics and searches for each of them simultaneously — Google calls this "query fan-out" — then merges everything into a single response with follow-up capability. You can ask it things with text, voice, images, PDFs, or by pasting in a URL, and it retains history so you can pick up a previous line of questioning. Google has also rolled out a more advanced "Pro" model option inside AI Mode (built on Gemini 3 Pro) for harder, more reasoning-heavy queries, alongside a faster default model for everyday questions.
There's also a personalisation layer worth knowing about: if a user is signed in and has connected apps like Gmail, Calendar, or Google Photos, AI Mode can draw on that context — Google calls this "Personal Intelligence" — to tailor responses. A user can also explicitly tell AI Mode to remember or forget specific details. This matters for marketers because it means two people searching the exact same phrase can now get meaningfully different AI-generated answers depending on their own connected data.
Both systems are still pulling from indexed web content to generate their answers. That's the important bit for anyone doing SEO: the AI layer isn't inventing information from nowhere, it's synthesising and summarising what's already crawlable. Which means the fundamentals — crawlability, structured data, clear entity signals, genuine topical authority — haven't become obsolete. If anything, they've become the literal raw material these systems are working from.
Why this is changing how people search in the first place
Google's own data points to a genuine behavioural shift, not just a product push. Search interest in things like "how to digital detox" has reportedly risen sharply over the past year, and Google has been actively marketing AI Mode around helping people do more offline — booking classes, finding gear, planning events — treating the AI layer as a personal assistant rather than a directory. Whether or not that framing lands with users long-term, it signals where Google thinks the puck is going: fewer standalone searches, more extended conversations that resolve a task in one sitting rather than sending someone off to browse five different websites.
That's the crux of the disruption risk. A traditional search sent a user to your site to find the answer. An AI Overview or AI Mode conversation can resolve the entire query without a click ever happening.
Industry commentators, including SEO analysts quoted in trade press, have warned this could meaningfully cut into the organic traffic that publishers and small businesses have depended on for years — and we've seen early signs of exactly that in client data, particularly on broad, informational queries that used to reliably send traffic through to a blog post or guide.
What we're actually doing differently for clients
This is the part that matters most: tactics, not headlines. Since AI-generated answers started eating into traditional click-through rates, we've shifted a meaningful chunk of our SEO work toward what's often called AEO (answer engine optimisation) or GEO (generative engine optimisation). In practice, that's meant:
Rebuilding pages around direct, extractable answers. AI systems tend to favour content that states things clearly and early, rather than pages that bury the answer under a long preamble. We've been rewriting service and category pages to lead with a clear, concise answer to the likely query, then expand into supporting detail — the opposite of the old "keyword-stuffed intro paragraph" approach.
Adding proper FAQ structure and schema markup. Structured FAQ blocks with correct schema give both traditional crawlers and the AI summarisation layer an unambiguous read on what a page answers. We've rolled this out across service pages, product collection pages, and blog content for multiple clients this year.
Tightening E-E-A-T signals. Experience, Expertise, Authoritativeness, and Trust have always mattered to Google's quality systems, but they matter more now because AI Overviews and AI Mode lean directly on those same underlying signals to decide what's citation-worthy. That means real, named author attribution, clearer sourcing and citations within content, and cutting thin or duplicated pages that dilute topical authority.
Being more aggressive about technical health. An AI system that fans a single query out across dozens of sources at once has even less patience for a slow, broken, or badly structured site than a human searcher does. Site speed, clean crawl paths, and correct canonicalisation have moved further up our priority list as a result.
Watching citation patterns, not just rankings. We've started tracking which client pages actually get surfaced or linked within AI Overviews and AI Mode responses, not just where they rank in the traditional results. It's a newer, messier signal than classic rank tracking, but it's becoming just as important a diagnostic.
None of this is a silver bullet, and anyone promising a guaranteed way to "rank" inside an AI Overview is overselling what's currently a fairly opaque, constantly shifting system. But it's the difference between quietly becoming one of the sources an AI summary draws from, and disappearing from the answer entirely.
Should you actually be worried about this?
It depends heavily on what kind of traffic your business relies on.
If a meaningful share of your traffic comes from broad, informational, top-of-funnel search terms — "how does X work," "best way to do Y" — pay close attention. Those are exactly the query types most likely to be fully answered inside an AI Overview with no click required, and we've seen the sharpest drop-offs in exactly this category across client accounts.
If your traffic leans transactional or brand-specific — people already searching for your company by name, or actively ready to buy a specific product — the impact so far has been comparatively mild, though that's shifting too. AI Mode has been expanding into shopping and local availability, including the ability to check nearby stock and even contact stores directly, which starts to bring transactional and local search into scope as well.
It's also worth remembering Google has built explicit feedback loops into both products, and is upfront that AI-generated answers can contain mistakes. That tells you this system is still actively being tuned, not locked in. Which is genuinely useful information for anyone doing SEO right now — the sources these systems favour today aren't fixed, and there's still real, live opportunity to become one of them if your content is structured in a way these systems can actually read and trust.
A short glossary, since the terminology gets thrown around loosely
- AI Overview — the AI-generated summary box that can appear above traditional results on a standard Google search.
- AI Mode — a fuller, conversational search experience with follow-up questions, file/image input, and query fan-out.
- Query fan-out — the technique AI Mode uses to split one question into several subtopics, search each simultaneously, and merge the results.
- AEO / GEO — Answer Engine Optimisation / Generative Engine Optimisation, the emerging discipline of structuring content to be cited by AI-generated answers, rather than purely to rank in traditional blue-link results.
- E-E-A-T — Experience, Expertise, Authoritativeness, Trust — Google's long-standing quality framework, now doing double duty as an input to what AI systems treat as citable.
Google AI Search takeaway
Google's search bar is becoming an AI product first and a links directory second. That's a genuine structural shift, not a passing trend — and it's already changing how we brief content, structure pages, and report results for clients month to month. The businesses that come out ahead of this won't be the ones panicking about traffic loss; they'll be the ones treating structured, genuinely authoritative content as infrastructure rather than an afterthought.
If you're wondering whether your site is currently set up to survive — or actually benefit from — this shift, that's exactly the kind of audit our team runs day to day. Get in touch and we'll show you where you currently stand.
Categories:
General |
