What a Vibe Coded App Should Tell Publishers About User Needs

A few weeks ago, Jodie Hopperton, INMA's Product Initiative Lead, saw something on Instagram that stopped her. A friend had built her own personalized news product using Claude Code - not a prototype, but something she actually uses daily. She had a Financial Times subscription, couldn't find what she needed in the FT's own product, and solved the problem herself in about 30 minutes. She is not a developer. She had never paid for news before. She subscribed to a brand she trusted, then built what the publisher hadn't built for her.

Hopperton sent the screenshots to her FT colleagues and wrote a piece for INMA asking the industry a direct question: Why isn't news doing this? Hopperton concludes with a question about product/market fit: "How fast can we build it?”

This anecdote got me thinking. Why is it that legacy news companies so often seem to be behind in adapting to shifting audience behavior - whether that's reacting to the rise of the mobile phone as the dominant interface or social platforms or AI chatbots enabling users to get answers without interacting directly with media outlets. I think the answer runs deeper than just the speed with which media companies are able to build shiny new things with emerging technologies.

The underlying problem is that for the longest time advertising-financed legacy media didn't really have to put user needs first. That was definitely true in the media as gatekeeper age but it has carried over because putting user needs first is hard. It might mean giving up what has worked over decades and is deeply ingrained in the organizations' culture. And that seems to be especially true for media organizations in the German speaking world (called DACH countries for the country signifiers D, A and CH on their automobile license plates).

The pattern shows up consistently across markets, but it hits hardest where the organizational culture reinforces caution over experimentation. There's a name for that culture. Dmitry Shishkin, who has worked at the BBC, at Ringier, and now consults with media executives across markets, calls it governance-led - as opposed to product-led. I reached out to him via email. His read: "The most significant difference within the 'AI adoption in media' scene is structural; it's not related to pace, and it's hard to generalize across regions. It concerns how AI is positioned within organizations - governance-led versus product-led." He adds: "The most successful media organizations view AI as a paradigm shift, not a cost-saving tool with governance overhead.”

Governance-led organizations ask: How do we manage this? Product-led organizations ask: What do our readers need, and how do we build it? In the German-speaking market, the governance answer has dominated - and there are specific historical reasons why.

DACH publishers operated in a market with structural characteristics that made decline feel manageable longer than it was. Germany has Europe's largest newspaper market by volume - a dense local landscape of subscription papers, home-delivered, deeply habitual, rooted in communities in a way that has no real equivalent in the US or UK. German daily circulation has fallen from 27.3 million in 1991 to 10.2 million in 2024 - a 63% decline - but the base was large enough, and the subscription relationships sticky enough, that publishers could keep moving their existing model into new containers without ever confronting the real question: What do readers actually need from us?

Scandinavia handled this differently. Norwegian and Swedish publishers pivoted earlier and more decisively to digital subscriptions, but the deeper difference is that they never handed their audience relationship over to platforms in the first place. Because readers paid directly and came directly, publishers kept learning what those readers actually needed. 

By 2016, 22% of Swedish households had digital newspaper subscriptions, and overall daily readership stabilized at around 70% because digital subscriptions filled the gap left by print. In DACH, platform dependency interrupted that learning loop. Publishers optimized for reach instead of relationship, and the understanding of what readers actually needed atrophied. By the time AI arrived, Scandinavian publishers had years of direct reader data and habits to build on. DACH publishers generally did not.

I know what that question looks like when it goes unasked. I worked in a local German newsroom in the early 1990s. By far the largest category of our stories was reactive: What's on the calendar, and therefore what do we need to cover? The press conference, the council meeting, the annual report. The question that rarely got asked: Regardless of the calendar, what are the real problems in this community? What do people here actually care about, and what is our role in serving them?

That question - and the commercial logic that follows from it - is what publishers like Lookout Media in the U.S. and Village Media in Canada built their businesses around. Both started from a clean slate in terms of legacy traditions and they placed a specific community question front and center: What do the people in this place actually need to know, and who is telling them? They built editorial models around answering that question directly, covering what their communities genuinely cared about rather than what was scheduled. Both have sustainable for-profit businesses. 

As far as I know, there is no comparable publisher in the German-speaking market. I have been analyzing the media industry in the U.S. and Europe for fifteen years. If something like Lookout or Village Media existed in DACH, I would expect to have heard about it. I may be wrong - and if I am, I want to hear about it.

When digital arrived in the early 2000s, DACH publishers made the same mistake most publishers made: They put their content online for free, hoping mass reach would translate into advertising revenue. It didn't. What it produced instead was an optimization for clicks rather than for reader value - clickbait as the logical endpoint of a model that measured reach instead of relationship. That was not a technology failure. It was the user needs failure becoming more visible and more costly.

The print buffer that cushioned publishers through that period is now gone or shrinking fast and the financial pressure is only growing. The answer most publishers in the German-speaking market have reached for is AI. But it is largely framed as a cost-cutting solution rather than a framework to strengthen the core mission of journalism and better serve user needs. Anita Zielina, founder of Better Leaders Lab and a board member at Mediahuis and Funke, told me in an interview for trade magazine Kress Pro: "In the German-speaking market, AI is currently still treated primarily as an efficiency question.” That is the governance answer to a product question. It produces cost savings. It does not produce what the FT subscriber built for herself in 30 minutes.

That young FT subscriber told Hopperton: "Could my mom do it? No, not at the moment. But in a few years,  she could."The tools are improving faster than most news organizations are moving. Vibe coding - building functional software through iterative conversation with AI, no programming knowledge required - is already accessible to non-developers with patience and a clear sense of what they want. What took one technically curious FT subscriber 30 minutes today will take less effort in two years, and will be available to a much larger group of people. The gap between what publishers offer and what readers need is not shrinking. It is becoming easier to route around.

The question Hopperton put to the FT's product team is the same question all publishers face - do they understand what their readers need?