🟣 RIP Blue Links

Google made it official. But the traffic was already leaving — and the more important question is what kind of traffic it actually was.

Tony Napoleone

Tony Napoleone

— 5 min read
Share
🟣 RIP Blue Links

Publishers have a history with Google announcements. Take them seriously, watch the deadline slip, and eventually conclude it's probably not going to happen.

We did it with third-party cookie deprecation. Google announced the plan in 2019. Deadlines came and went — 2022, then 2023, then 2024 — until Google officially abandoned it in April 2025. Six years of "this is definitely happening" before most publishers stopped planning around it entirely. The threat was real. The urgency kept fading. People got comfortable.

The same thing played out with AI search. The writing has been on the wall for at least two years. AI Overviews started rolling out in 2024. Zero-click results were already climbing. Reports came out every few months showing organic search traffic declining for publishers. And yet most organizations didn't fundamentally change their audience development strategy. The bet — mostly unconscious — was that Google would keep the blue links. That the core bargain between search engine and publisher would hold.

At I/O in May, Google announced what it called the biggest upgrade to Search in over 25 years. Blue links aren't gone, but they've been demoted. AI Overviews and AI Mode, powered by Gemini, now sit at the top. Zero-click answers get top billing. Google called this empowering for users. For publishers, it's the official end of a distribution model.

The backlash was fast. DuckDuckGo reported app installs up 30.5% on a single day after the announcement, with sustained growth through the Memorial Day weekend. CEO Gabriel Weinberg put it directly: "Google is force-feeding AI with no way to opt out." The frustration is real. But DuckDuckGo holds about 2% of the U.S. search market. The protest vote exists. It won't move your referral traffic.

Here's the honest context that's easy to skip over: for most publishers, this traffic was already disappearing. Organic search for media companies has been declining for 18 to 24 months. The I/O announcement isn't the moment this started — it's the moment it became official. Which means the pain is real, but the shock probably shouldn't be.

And it raises a more useful question: what traffic were you actually losing?

Because a lot of organic search traffic was never really your audience. It was someone who Googled a question, landed on your page, read one paragraph, and bounced. It propped up your pageview counts and gave your display campaigns something to run against — but those weren't your buyers. They weren't going to subscribe. They were search tourists, and the AI is eating them. Google's own quality raters guidelines were never actually a ranking rulebook to begin with — human raters evaluate whether the algorithm is working, not what gets served — so even the publishers optimizing hard for quality signals were working with a softer lever than they realized.

So before the panic sets in: is this an audience development problem, or is it an ad impression problem? Those require completely different responses. If what's disappearing is anonymous volume that was making your display numbers look better than your actual audience warranted, then what you're losing is a distortion. Getting clear on that is useful.

That clarity also points to where the real work is. Only 24% of Americans had high confidence in big tech companies in 2025, down from 32% in 2020. Users are skeptical of the platforms and increasingly looking for alternatives — but most of them aren't going to find a better one. What they will do is develop a stronger pull toward sources they already trust. That's the opening for publishers who have been building direct relationships with their readers.

Here's what the shift actually demands from an audience standpoint.

Free organic volume is going away. Every page visit now has to work harder. The publishers who win in this environment are the ones who treat known and unknown visitors completely differently — who can recognize a first-time referral visitor versus someone who has been back six times in the past month, and give each of them a fundamentally different experience. Think a metered experience for the first visit, a direct subscription ask by visit four. The repeat visitor — the person who keeps coming back because you have long-tail research or expertise nobody else has — isn't going anywhere. But conversion is now the most important metric in your stack, not traffic volume.

That means building systems to move unknown visitors toward a relationship — registration walls, email capture, progressive profiling — before they leave the same way they arrived. Every unknown visitor is a conversion opportunity, and you need a system to move them through that funnel rather than let them leave the same way they arrived.

On the content side, this is where AEO and GEO — answer engine optimization and generative engine optimization — become real work rather than buzzword territory. Be skeptical of vendors promising they can get you consistently cited as an LLM source. The landscape is too early and too fluid for that guarantee. But the underlying work is legitimate and most of it follows SEO best practices you already know: structured data, clear topical authority, well-organized information that AI systems can parse and cite.

The key is being selective about where you put that effort. Not every page on your site deserves deep optimization for AI referral. A restated press release with no original reporting is not going to be reliably cited by an LLM, and spending resources trying to make it one is a waste. But there are key pages on every publisher site that stand a real chance of being pulled into AI-generated answers — your annual benchmark data, your original research, the topic hub where you've published 50 pieces over five years and genuinely own the category. Those are the pages to focus on. Make sure the top-line information is structured cleanly, that the data is easy to pull, and that the page signals authority clearly. That's where the AEO work pays off.

And here's the part of this that should actually give you some optimism: the traffic that does come through AI referrals is highly qualified. You might go from a thousand visits on a page down to fifty coming in from an LLM source — but those fifty are far more likely to take action. They're not browsing. They arrived with intent, got pointed to you as a credible source, and they're researching something specific. For subscription conversion, lead generation, webinar registration, white paper downloads, event attendance — that visitor converts at a much higher rate than the search tourist ever did.

The dashboards are going to look worse. Traffic numbers will drop, and in organizations where pageviews are still the primary currency that will be uncomfortable to explain. But the underlying question for audience developers isn't how to get the traffic back.

It's whether you've been building for the visitors who were worth keeping in the first place.

Sources

Read more