An executive at a major search company recently told me something blunt: News publishers should expect zero to near-zero traffic from search in the next two to three years.

It landed with me harder than I expected — not because it was shocking, but because it felt so obviously true.

And the data is starting to show it.

But product logic tells us that answer engines — especially AI-powered ones — are designed to keep users where they are. They’re optimised to give complete, follow-up-ready answers without ever sending someone off-platform.

And why would they? Clicking out breaks the flow. It’s friction. It’s bad UX.

We’re already seeing this shift formalised through AI licensing deals — ones that focus on display, not referral. It’s not about sending users to you; it’s about showing your content there.

This is Google’s innovator’s dilemma: Should they risk their lucrative search business by leaning fully into AI overviews?

But here’s the thing: We’re facing our own version of that dilemma in news media.

We have to serve loyal users who know and trust our interfaces — the apps, Web sites, and e-readers they’ve grown familiar with. At the same time, we’re being challenged to show up in entirely new environments: chatbots, smart assistants, answer boxes. These places have none of the navigational context we’ve spent decades perfecting.

And history isn’t exactly on our side.

When the Internet first disrupted news, we gave content away for free. It took years to unlearn that. Then came social media. We still haven’t figured out a stable business model for that ecosystem. With each new platform, we risk cannibalising the very business we’re trying to protect.

This time, the shift isn’t just about format — it’s about function. Generative AI is setting new user expectations, and tech companies are defining the standard.

At the Product & Tech Seminar at INMA’s World Congress last month, Varun Shetty made a fairly compelling case that for retrieval-augmented generation (RAG) content, links will still matter. These models won’t have the full context of a news story, and in many cases, they will need to direct users to publisher sites. 

The challenge for us lies in bridging that moment — from a clean, highly personalised UX in an AI environment to a potentially cluttered news Web site. This is a design, product, and editorial challenge all rolled into one.

If you’re in product and not using AI tools regularly, you’re behind. Now is the time to experiment — not just as users, but as builders. We need to understand how to translate our value, content, and business models into an AI-native world.

Because the question isn’t whether traffic from search will disappear.

It’s: What are we building next? (as you can read in INMA’s April report “As Search Ends, Here is What’s Next.”)

That’s exactly what we’re trying to solve through INMA’s new partnership with OpenAI — a series of member roundtables, prototyping credits, and a forthcoming report focused on shaping the next-generation user journey. Read more here.  

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .