Re:filtered #31: What counts now

Greetings from Vienna, and welcome to the 31st edition of this monthly newsletter on civic media in a moment of systemic disruption.

Some housekeeping: we're redoing how we run our newsletters:

• re:filtered (this one!): monthly long-reads on media strategy in this moment of disruption. Read it if you want big-picture orientation.
• re:noted: restarting our biweekly field notes as nuggets of inspiration that service design can bring to journalism, by Madison Karas and me. Read these if you want practical ideas to try in your own work.
• tbd/con: updates on our virtual, interdisciplinary gathering on AI and information in September, by Matthew Gore-Kormanik, David Kuszmar, Madison, and me. Read these if you plan to join some unconventional conversations on AI. We still accept pitches, and you can register to attend for free.

Nothing changes for you. You can un/subscribe to each and all here, or whenever you visit gazzetta.xyz.

---

We're picking up where last month ended: if people largely no longer reach journalism by searching and clicking, what should we measure instead to understand resonance?

The industry's measurement problem is older than AI. A craft has no definition of success until you tie it to an intended outcome; like the samurai's sword, journalism has served everything from security to oppression (or a restaurant: from nutrition to food poisoning).

For the past century or so, many practitioners' intent has been civic: someone monitors the exercise of power so that community members can navigate it better, and those in power know they're being watched.

But have we measured ourselves by that claim? We measured the occasional full-stack success (#metoo, Watergate), output (the investigation), and above all consumption, which draws a straight line to money: circulation, page views, engaged minutes, subscriptions, ads.

For two decades the measurement line went up on its own as the internet filled up with addiction-optimized products, and the mission was assumed to be going up with it. That era is over.

What survives of the dashboard

The most disciplined version of consumption measurement from that era is a formula that publishers, most famously the Financial Times, borrowed from customer analytics, called RFV: recency, frequency, volume.

It tracked when someone last visited, how often, and how much time they stayed while there. It was built for a world where all three ultimately happened on your own site. It needs an update.

Volume now tells you little, and less every day. Recall the two modes of information acquisition: the hunter who comes once for one specific, deep answer, and the forager who grazes for unexpected bliss. They produce opposite volume numbers, and the smaller one is often the healthier relationship.

Patrick Radden Keefe of The New Yorker recently told a story about a friend who loves the magazine, subscribes, supports it, and feeds every ten-thousand-word piece into ChatGPT for a summary. Keefe's answer: you're using it wrong. The pleasures are literary; "I want to seduce you with the narrative," he said.

The two modes explain why this is happening: The friend is hunting, and for him the summary is the service. Keefe writes for foragers, and for them the experience is the service: the voice, the detour, the unexpected bliss no summary can carry. A single story cannot choose which reader it gets.

A publication can know which mode it serves, promise it, and measure it. For work like Keefe's, time spent inside the experience is not a proxy for the outcome. It is the outcome, the way a dinner is.

Recency used to rank loyalty. Read it now more modestly as opportunity: how recently someone engaged tells you how open the window is, when to invite an action, when a relationship is going cold.

Frequency is the sole pure survivor. It is the closest a consumption metric comes to measuring a relationship.

Someone who returns every week, unprompted, because it is the weekend and the brief always lands on Saturday, is showing you a habit.

A visitor sent once by a search engine (many) or chatbot (few) is a prospect, not an audience, whatever the dashboard says.

That disconnect is one reason major newsrooms are now (often quietly) shrinking the audience teams they built over the past decade to chase search and social traffic, or shifting staff to product roles.

The total-reach trap

Which brings me to a response in the nonprofit media sector that I find fatally shortsighted: total-reach reports across all platforms. Take the YouTube views, add the TikTok impressions, the newsletter subscribers, the site sessions, and hand funders one ginormous number.

In my conversations I hear more newsrooms turning to total reach numbers as they compete for a shrinking philanthropic pot. It sets them up to fail.

Success always depends on who the activity is for. At a workshop I ran last week, most of the room of newsroom executives said they explain their impact first to donors, not to the people they serve. The total-reach number is that habit in its purest form. It impresses donors, and it gets gamed in perverse ways.

Some of the things I have seen "public interest" newsrooms do are mind-boggling: metadata designed to catch pedophilic search queries, double pop-up links (site opens twice!), server farms in war zones that run scripts to play YouTube videos, pop-up ads bought by the million.

A more benign example: I worked with a newsroom that proudly reported more followers across platforms than the country it covered has inhabitants, and genuinely presented this as a proxy for success.

The aggregate number conceals concentration risk: a disproportionate share of those views sits on platforms whose numbers swing for reasons entirely outside the organization's control. One algorithm change, and the reported impact halves overnight while the work stays the same.

A metric you can lose to someone else's product decision is nothing to build a strategy on.

Now, two things are true at once:

1. most information consumption really does happen on those platforms (and chatbots);
2. and our strategic success metrics cannot be defined by them.

Platforms are mere touch points, moments where the work brushes against someone's day. They are not the service (unless we are hired to be messengers of a desired change or cause).

Success is what an information service changes for people, rooted in their own agency, and it is measured in the far less bombastic places where the relationship actually lives.

Lifetime civic value, revisited

So where to look for validation when dashboards with their artificial precision disappoint? Start with what a newsroom can honestly claim.

No funder or investor can expect systems change from a newsroom, yet they so often do; it only invites hubris and false claims, a convenient camouflage for purely personal short-term ambitions.

What you can show is who was equipped to act, and what they then did. That claim is precise enough to fail, which is what makes it worth something, and has social value.

"Strengthen democracy" or "speak truth to power" can never fail, which is why it convinces few and rarely achieves anything.

Two years ago, I proposed lifetime civic value as one way out. Businesses, including many publishers, value a customer by multiplying the revenue they bring per year by the years they stay. As a thought experiment, I suggested replacing revenue in this equation with civic action: count the specific, observable things the work enabled people to do, and multiply by how long the relationship lasts.

A few newsrooms discussed it at the time but none adopted it because it would have pinned them to a claim that can fail, and funders were at peak dashboard craze with no appetite for non-comparable metrics.

For both sides, the views-and-likes slot machine felt safer than a model that holds you accountable to an outcome.

This is one recipe, not the recipe. Any measure built on observable actions inside relationships you own does the same job; I'm just mentioning it to say that alternatives to fluff and slot machines can exist, if we want them.

Two years later, we increasingly need them. The machines are more visibly rigged, their owners more corrupt, and the payouts smaller.

Stepping away from this Faustian bargain is becoming less a choice than a necessity.

This can't be solved with better content or stickier products. It is a leadership and strategy problem. We’ve been calling it a service problem.

Many working journalists would be relieved: their work becomes honestly measurable and starts to feel a little bit more real instead of "feeding the beast."

Not "we reached 1.2 million people" but claims a team can stand behind: 23 parents used our guide to find a school place this fall; readers recovered thousands in overpaid fees after our explainer, and wrote to tell us; not gazillions, but each one a real person whose life improved, and each either happened or it did not.

That is what a funder can trust, what a team can rally around, and what no algorithm change can take away.

The consumption metrics then fall into place as supporting cast: frequency says the relationship is alive; recency says when to invite the action; the action says the work mattered.

In such a world, audience and community teams could do so much more than being embedded marketers.

Three success stories, reconsidered

Three moments from recent workshops, anonymized.

One team shared a story about members of a vulnerable group, people who often feel unsafe and alone, helping each other get through pressure from local authorities, and readers reposted it into community groups, praising the protagonists' courage.

They were apologetic, their instinct was to file that as soft impact because it didn’t go viral or lead to any systemic change. That instinct is wrong twice over.

Seeing that mutual help works is the difference between isolation and the sense that community is possible, one of the most basic human needs there is. And reposting such a story into a community group can be a trust decision that carries personal risk; sharing under risk is among the strongest signals a newsroom can earn.

The open question is what turns one great story into a promise, say a recurring strand of how people handled it, so the claim becomes "people in trouble know they can come to us and find out what worked for others." Then watch who returns and who shares you a second time, because that second share vouches for you, not for the story.

Another team shared a dream case: a reader trusted them with a risky tip about an institution about to put people in danger, they verified and published, and thirty minutes later that institution rescinded its plans. It deserves real admiration, but again as an impact narrative it stands on rented ground.

Institutions rarely respond because they want to be useful; they respond when responding is the cheapest way to keep things quiet, and that lever can be withdrawn the day ignoring you becomes cheaper: it’s the autocratic cousin of the algorithm change.

What nobody can withdraw is what these affected people got directly (before we talk about the institution's reaction):

They felt heard, learned they were not alone in finding it wrong, and saw that speaking up through this newsroom is survivable, a trust that shows up as behavior: the forward, the return, the next tip arriving.

The test I encouraged them to sit with: if the authorities stopped reacting tomorrow, what would readers say they still get? That answer is the service.

In the real world after the end of the illusion of the end of history, policy reversals cannot be a realistic baseline expectation for impact.

It’d be like defining a restaurant by the business deals closed or marriage proposals accepted among its diners. A bit of a stretch, however welcome.

Where the counting ends

The third moment came from a conversation afterwards, a pushback delivered as a metaphor: a heat warning has a clear, countable job, it sends me looking for water and shade.

Journalism about why the heat keeps getting worse has a different ambition: it wants to gather us around a shared problem. But something as abstract and difficult as the climate crisis that caused the heat does not set that table by itself.

The slow work of making it discussable resists counting, and on an ordinary day it can even feel like an annoyance rather than a service.

The participant was right. An operation built only on provable outcomes will starve the work that creates the conditions for outcomes.

Yet the difficulty can't be the excuse: calling something a wicked problem can conveniently assume an interest the audience may not have, and demand is not only found, it is also built.

If nobody wants the conversation we're having, the answer is to earn the appetite, not to claim credit for the annoyance.

That leaves a test for a number on a dashboard: what decision does it help you make? If the honest answer is none, except to impress, it is donor-oriented decoration.

The lines worth watching tell you whether people can do something they could not do before, because you supported them functionally or emotionally, and whether they would come looking for you if you went quiet.

In service terms, Radden Keefe's promise is not any single story; it is seduction on schedule, the standing offer that an evening with his byline will be repaid. His readers prove it in the purest way: they follow him from the opioid crisis to Northern Ireland to Mexican cartels. They hired the storyteller, not the subject. Return that survives a topic change is the strongest relationship signal a dashboard can show, and no summary can take it from him.

And in an ideal world, everyone with his gift gets a shot; nobody rations storytellers. The better question is what the seduction is for. Keefe spends his making people care about opioid settlements and decades-old disappearances, subjects they never asked to care about.

The same craft can be pointed at critical information needs: the school system, the housing decision, the checkpoint, the heat. Can narrative seduction help people make more informed choices in their own lives? If public media needs a mandate worth defending, that might be it: spending the best of the craft on the information people need most, not only on the stories that sell themselves.

But the two overlap more than our org charts suggest. A trusted voice is what makes utility habitual, and proven utility is what earns a voice its trust. Most newsrooms will live closer to the useful end than the literary one, and that is fine. The mistake is being one while measuring yourself as the other.

In research terms, this should lead to a resurgence of qualitative work, but it requires an industry leadership willing to evolve.

We keep collecting methods in our freshly updated Service Desk resources, and a proper publication is in the works.

If you want a quick read on where your own organization stands, our five-minute gut check tells you whether people value you, need you, or just read you.

How do you measure your relevance and utility? I'd love to know. Respond to this email or message me on Signal at patrickb.01.

See you next month!

P.S. I set out to make this issue 50 percent shorter than the last. It fought back and landed at about 20. Whether even that improved the intended outcome of de-enshittifying the journalism support industrial complex, or only the volume metric, you'll have to tell me.