What the Best Newsrooms Know About Failing

The best product decision you’ll make this year might be cancelling something.

Not just pivoting or quietly deprioritizing it, but actually killing it and moving on. Many media organisations don’t have the option of failure built into their process and it’s holding them back.

Shared at the 2026 International Journalism Festival in Perugia, here are the tips from newsrooms who who are good at failing (in the best way possible).

Build the kill option in from day one

Most organisations try to meet their way out of misalignment, but meetings rarely bridge the gap between leadership’s vision and what actually gets executed on the ground. As Felicitas Carrique of the News Product Alliance put it bluntly: “We are not going to ‘meet ourselves’ out of this problem”.

So what does work? For Zetland, the solution is a forcing function: a concrete, cross-functional project that requires people to actually do something together. Their “brain trust” model pulls people from several different teams into the same room from day one (e.g., editorial, product, marketing/audience, visual storytelling)

The ideal behind bringing these groups together from the get-go is simple: If you can’t get everyone around the table genuinely excited about a project — or if team members can’t see how it fits their department’s mission — the project should probably die. Better to find that out in week one than month twelve!

The brain trust’s five-stage process

Zetland has used this brain trust model to develop their podcast suite. In this case, they used brain trust’s five-stage process to move an idea from pitch to full launch:

1. Pitch: The creator shares a short demo (e.g., five minutes of audio, a rough mockup) and the group reacts honestly to the core idea
2. Prototype: If greenlit, a rough version is built and reviewed together; the group decides whether to proceed or send it back
3. Design: Visual identity, sound, format (only once the idea has genuinely earned buy in)
4. Beta & KPIs: The team sets clear success metrics for the first three months before launch
5. The kill option: At every single stage, the brain trust can give a thumbs down to fill the project.

The key insight here is structural: by making “no” a legitimate outcome at every stage, Zetland removes the social pressure that keeps bad ideas alive. The kill decision doesn’t get more comfortable over time — it gets more expensive. Building that exit into the process from the start is what keeps failure cheap.

Getting a “resounding meh”? Buy a cake

When testing new features, the most dangerous result isn’t negative feedback, it’s indifference.

Olivier Baumann, CTO and Deputy Director of Republik Magazine, learned this firsthand when designing and launching a short-form content feed designed to solve users’ “time poverty.” Loyal subscribers said the design “looked nice.” That said, the data told a different story: a resounding meh, no measurable impact, and months of resources spent proving a hypothesis they could have killed much earlier.

Now on the other side of this experience, Olivier identifies three tips to help:

• Decide before you test: If your team isn’t genuinely prepared to kill the idea based on what you find, don’t bother running the test. User research shouldn’t be a rubber stamp.
• Start smaller: Instead of building for 15 months, release the simplest possible version that tests your core hypothesis. You’ll learn faster and lose less if you’re wrong.
• Define success before you launch: Without agreed criteria, a failing product can drift for months with nobody feeling authorised to stop it. Agreeing upfront on what “this worked” looks like is what makes the kill decision clean rather than painful

The pattern here is the same as Zetland’s: failure only becomes expensive when there’s no agreed mechanism for recognising it. When the data says stop, stop. And buy a cake (yes, really!). Treating a timely shutdown as a strategic win rather than an admission of defeat is what makes the next experiment feel safe enough to try. Republik went further, presenting a candid “blue-slide” post-mortem deck to the entire newsroom explaining exactly why their short-form feed failed.

Surface the hard decisions before they surface themselves

Even with the best ground-level processes, misalignment at the top will eventually undo them. If the C-suite isn’t resolving strategic tensions explicitly, those tensions don’t disappear — they just get pushed down, where nobody has the authority to make the call.

At The Atlantic, Executive Director of Product, Mariah Craddick’s solution is a structured ritual: a monthly Product Strategy Review with a deliberately uncomfortable section at the end. They call it “Spicy Topics.”

The meeting opens with ten minutes of silent reading with C-suite and senior leaders working through a shared draft document and adding comments directly onto it. What follows is a discussion driven by those comments.

This “spicy” section is saved for last, and for good reason. It’s where the real tensions live: acquisition versus retention, editorial priorities versus advertising, what the roadmap promises versus what the organisation can actually deliver. The explicit goal is to arrive at a decision point, a moment where leadership has to choose one path over another rather than let both limp forward in parallel.

This matters because competing strategies don’t just waste resources, they create the exact conditions that keep failing products on life support. When the CEO hasn’t explicitly chosen acquisition over retention, no product manager has the authority to sunset a newsletter that’s serving the wrong master. The “Spicy Topics” meeting is where that authority gets assigned.

The takeaway

The newsrooms getting this right aren’t the ones that never fail. Failure isn’t the problem. It’s expensive, late, unacknowledged failure that is.

The ones who are moving the needle are the ones that have built systems to fail on purpose — early, cheaply, and without shame. The organisations getting this right are building explicit moments of accountability early, into the open, and at the right level, into the process instead of hoping misalignment resolves itself.

So, build the kill option in from day one. Define what success looks like before you launch. And when the data says stop — stop.

Then buy a cake, and start the next experiment.