AI Overviews and chatbots have made visibility almost free, and made the click to your site rarer than it has ever been. For any publisher paid on visits, that gap is the problem to solve. Here is one route that's working, and the other avenues worth knowing about.
For most of search's life, being seen and being visited were the same event. Earn the ranking, and the click came bundled with it. Visibility was the difficult, valuable thing to win, and traffic followed it down the page. That bundle has come apart.
AI Overviews now sit at the top of a growing share of results, answering the question in place, and the latest data shows AIOs are triggered at a staggering 43% of searches on average. However, when they appear, people click through far less. Ahrefs measured the top-ranking page losing roughly a third of its click-through rate when an overview appeared in early 2025, widening toward half by the end of it, and that is if they get a click at all.
None of that is a visibility problem. Impressions are up, often even more so than before, while rankings hold. The thing that has gone scarce is the visit.
The difference between whether this is a crisis or a shrug depends entirely on how your business is paid. E-commerce and even some services are generally less impacted. Any site that was relying heavily on traffic itself for income, though, is now feeling penalised for its work.
While a subscription-led title can lose shallow, answer-shaped traffic and barely feel it, because the reader who converts was always going to scroll past the overview and click, a publisher paid by advertisers is paid in a different currency: it's paid for the audience it delivers. It is the sessions that put ads in front of eyes. For that model, AI Overviews drain the account in precisely the unit its revenue is measured in, the visit itself.
That's the account we'll use to make this concrete. Across two consecutive six-month periods, one publishing client saw every visibility metric improve while the visits fell. This wasn't a ranking failure. It was the decoupling, drawn in a single site's data.
| Metric (6 months on 6 months) | Before | After | Change |
|---|---|---|---|
| Google organic clicks | 157,938 | 152,715 | −3.3% |
| Google organic impressions | 18,040,468 | 21,060,380 | +16.7% |
| Organic CTR | 0.9% | 0.7% | −0.2pt |
| Average position | 11.5 | 9.2 | ▲ 2 positions |
| Google Discover clicks | 37 | 8,572 | ×232 |
| Google News clicks | 114 | 1,295 | ×11 |
| Total search-surface clicks | 158,089 | 162,582 | +2.8% |
Illustrative first-party Google Search Console data, single publishing account. Discover and News figures reflect the intervention described below.
Read the top rows alone and the account looks triumphant: impressions up nearly 17%, average position up two places. Read them as a business and the story inverts: click-through fell from 0.9% to 0.7%, because the brand was appearing more and being clicked less. More seen, less visited. That gap is the whole problem in one line.
Once you stop treating “traffic” as one thing, the modern landscape gets clearer. Each surface that can carry a publisher's content settles up in either visibility or visits, and only some of them pay in the currency an ad-funded or subscription-based business banks.
AI Overviews pay in visibility. You appear, sometimes you're cited, and the visit usually stays on Google's page. Valuable for awareness, close to worthless for sessions.
LLM referrals mostly pay in visibility too, for now. This is the avenue everyone is watching, and it's worth watching. But keep the scale honest. While the traffic they bring is useful and growing with adaptation, overall usage and sessions often account for around 1% of overall engagement, which brings them down the bottom end of channels to invest in. For organic, the issue remains the same: citations without visits are still largely unhelpful for the bottom line.
Discover and News still pay in visits. They surface content to people the moment it's relevant and send them to the site to read it. Plenty of publishers have worked these surfaces for years. What's changed is the stakes: when being seen in classic search no longer guarantees a visit, the surfaces that still deliver one stop being a nice-to-have and become the core of the plan.
Organic search still pays in visits, too, and it's still where conversions live. The point was never to abandon it. It was to stop expecting it to carry volume it can no longer carry alone.
This is one route, not the only one, but it's a route that produced real visits on a real account, and the logic travels.
It started with a large-scale gap analysis: consumer trends and search intent mapped by region, to find topic areas with genuine headroom. Each area was then split into two lanes, so a single topic could serve both the reader who wants it today and the one who'll want it in a year.
Discover and News reward speed and relevance. Most Discover traffic to a story lands in its first couple of days, then decays; covering a moment on day one can outperform covering it a week late several times over. Short-lived, but capable of moving serious volume while it’s live.
Content built to compound, but only if it's treated as a living asset: refreshed on a cycle, updated, and surfaced rather than buried under whatever published most recently. Evergreen only keeps earning if you refuse to let it sink.
Underneath the content, two things did the quiet structural work. Site architecture was reworked so Google could crawl, classify and resurface the right pages onto the right surfaces, coupled with the correct schema markup so everything is given for a seamless SERP generation. Authority signals were built deliberately: the E-E-A-T groundwork that increasingly decides whether Discover and AI surfaces treat you as a safe, citable source at all, rather than merely a page that exists.
Running through all of it: freshness, which is more of a signal now than it has ever been. Google ran its first-ever Discover-specific core update in February 2026, leaning harder into topic authority, expertise and timeliness. So freshness wasn't a tactic bolted onto one lane, it was engineered into the whole model. News was built to be first, evergreen was built to stay current. Done consistently, that raises the trust the whole domain is read with, and a site that reads as alive earns lift across every surface at once, not just the page you updated.
Once the architecture, schema, freshness and trending coverage landed, a surface that had been sending this account very little started sending real, engaged visits.
Classic organic gave back 5,223 clicks to AI Overviews. Discover and News returned 9,716, close to twice the loss, and delivered them as readers who chose the story from a feed rather than typed a query and got captured by an answer box. Total visits across Google surfaces didn’t just recover. They finished ahead.
It's tempting to read AI Overviews as the end of search traffic for publishers. The more useful reading is narrower: search stopped bundling the visit with the visibility, and the two now have to be won separately, sometimes on entirely different surfaces.
The traffic didn't leave the internet, it just changed its routes. Which means the job isn't to mourn the click AI Overviews took. It's to work out which surfaces still pay in visits and build momentum there. The publishers who come out ahead will be the ones who can adapt to new challenges, and get their content seen by the avenues that still drive valuable traffic.
This is what giving a damn looks like in practice. Tell us what's not working.