…Understanding AI search is something we are learning…
For years, the pattern held: add good content regularly, and more people find your site.
Traffic would flatten out a little once a site was well established, but the relationship between fresh, well-structured content and visitor numbers was reliable. That pattern hasn’t broken, but the way it shows up in analytics has changed, and it’s worth understanding why. It’s also part of why we recommend WordPress rather than newer builders like Webflow, Framer, Wix, or Shopify. WordPress’s open, plugin-driven structure gives us the deepest level of control that both traditional search and AI systems rely on. And the adage that content is king still holds.
It’s not just the website itself, either. Posting regularly on your business’s social channels, and having people on your own team (or entourage) share and engage from their personal profiles too, helps signal that a business is active and genuinely engaged with its audience, which feeds into how visible and trustworthy it appears more broadly. Your Google Business Profile continues to be important: a complete, accurate listing on Google Maps with a steady flow of genuine positive reviews is one of the clearest trust signals a local business can have, and it plays into local search and AI-generated answers.
What’s actually changed
Search itself hasn’t gone away. Google is still, by a wide margin, where most people start looking for information and it’s not being replaced any time soon. What’s changed is that Google (and a growing list of AI chatbots like ChatGPT, Gemini, Claude, Copilot) now often answers the question directly, on the results page or inside a chat window, without the person needing to click through to a website at all.
That’s the shift behind the phrase you might be hearing more of: “zero-click search.” The content is still doing its job. It’s being read, trusted enough to summarise, and used to answer someone’s question. It just doesn’t always register as a “visit” any more.
Why Google Search Console only shows this from 18 May 2026
Google Search Console added a dedicated report in June 2026 that separates out how often your pages appear inside Google’s AI Overviews and AI Mode. Before this, that activity existed but was buried inside your general data with no way to isolate it.
Two things are worth knowing about this new report so the numbers aren’t over- or under-read:
- It doesn’t backfill. The report only has data from 18 May 2026 onward. There’s no way to see whether your site was appearing in AI Overviews last year, or last month, the “before” picture simply doesn’t exist.
- It only shows impressions, not clicks or the actual search queries involved. You can see that a page appeared in an AI answer, and roughly how often, but not what someone searched to get there or whether they clicked through. It’s not a full report.
- And not every website has this report populated yet, at least for now.
…Understanding AI search…
Search is still the dominant channel and will be for some time, but a meaningful and growing slice of how people find information has already moved into AI interfaces, and that slice is currently very hard for anyone to measure directly.
Why Google analytics hasn’t caught up
Google Analytics was built to measure visits, sessions, and clicks, that happen once someone lands on a page. It wasn’t designed to measure what happens before that: whether an AI system read your content, judged it trustworthy, and used it to answer someone’s question without ever sending that person to your site.
That’s not a flaw in the tools, it’s a gap in what they were built to do. It means a genuinely healthy, well-performing piece of content can show flat or falling visit numbers in Site Kit while still doing exactly what it’s meant to do, just somewhere your analytics can’t see.
The bigger blind spot: it’s not just Google
Here’s the part that’s easy to miss: Google Search Console can only ever tell you about Google’s own AI features. It has no visibility into ChatGPT, Claude and other A.I.s, these are separate companies which means no data for Google. If your content is being cited in a ChatGPT answer, there’s currently no report anywhere that will tell you that happened.
A handful of specialist tools are emerging to try to close this gap, some free and open-source, some paid subscriptions, that periodically test real questions against multiple AI platforms and log whether a brand gets mentioned or cited. This space is young and the tools are still maturing, so for now the most reliable (if manual) method is simply asking the kinds of questions your customers would ask, across a few different AI tools, and checking by hand whether your site or business gets a mention.
Why good SEO still matters, arguably more than ever
This is the reassuring part. The fundamentals of a well-built website such as proper heading structure, descriptive text on images, a clear page and post hierarchy, internal linking that helps both people and search engines understand how your content relates together, fresh and regularly updated content, and genuine backlinks from other credible sites, aren’t being made obsolete by AI search. They’re becoming the raw material AI systems use to decide what’s trustworthy enough to cite.
Google and AI tools don’t invent answers from nothing. They read, rank, and summarise real content from real websites, and they lean on the same signals search engines always have to judge quality and relevance: structure, freshness, authority, and clarity. A site that’s well optimised for traditional search is, in effect, already most of the way toward being well optimised for AI search too. Freshness in particular seems to carry real weight, so content that’s been updated recently appears meaningfully more likely to get cited than content left untouched for years.
In short: everything we’ve been doing such as the regular posting, the proper page structure, the earned backlinks, was never wasted effort, and it isn’t wasted now. It’s arguably doing double duty.
Google is still dominant…for now
It’s worth keeping this in perspective. Traditional search still accounts for the large majority of how people find information online, and Google alone still handles trillions of searches a year. AI chatbots are growing quickly and taking on a real, growing share of research-type and advice-type queries, but they haven’t replaced search — most people using AI chatbots are also still using Google regularly, often to double-check or dig further.
The realistic read is: search is still the dominant channel and will be for some time, but a meaningful and growing slice of how people find information has already moved into AI interfaces, and that slice is currently very hard for anyone to measure directly.
The reframe
If a websites data graph looks flat or slightly down, that’s a fair thing to notice, but it isn’t necessarily evidence the work has stopped paying off.
It might be that:
- The content is still being found and used, just via an AI summary instead of a click-through. A “visit” that never registers as a visit.
- The site is appearing in places we genuinely can’t measure yet, particularly outside Google’s own ecosystem.
- Traditional analytics were only ever built to see one part of the picture, and that part is shrinking as a proportion of the whole, even while the content’s actual reach may be holding steady or growing.
None of this means measurement doesn’t matter, it means the measurement tools are still catching up to where the behaviour already is. Where we can get real evidence (like Google’s new AI Overview impressions data), we’ll track it and report on it. Where we can’t yet, the honest answer is that a gap exists, and the best current defence is the same one that’s always worked: build the kind of website that’s easy for people and machines alike to trust, understand, and want to cite.
Code: COMP.DISCOVERY
