Let us go, you and I, as we walk together towards your ice cream shop down the street. We’ll get there soon, but bear with me as I build my case — and our appetites.

Your audience is only one kind of user. What I mean by this is that a user who consumes your content in a relatively passive audience-like way is only one kind of user.

You have other users who need your content in active formats other than journalism — reports, papers, events, briefings, databases, calculators, algorithms, curricula — as tools or inputs or services for their work. These are policy people in global capitals, NGO workers, other journalists, government officials, bureaucrats, think tank types in Oslo, big tech researchers from a MANGO (Microsoft, Anthropic, Nvidia, Google, OpenAI) subsidiary, teachers looking for contextual teaching materials, or a rural community development council in central India or Guatemala. Call them B2B users, if you like — they consume your content, but not in the way that an audience does. They need your content in order to make better decisions at their work, but in different formats that makes their usage more effective.

You also have another, non-human, user — AI. It’s the A in your B2A2C usage pattern. The existence of any agentic AI — or AI as an intermediary — means you’re also producing your content for machines (whether or not they’ve asked your permission). Anything you publish has a solid chance of being ingested by an LLM at some point. Either way, you might as well make it machine-readable if you aim to own any part of the conversation in the next few years.

Oh, look, here we are. Look at all these flavors. We should try a few. What are you feeling like this afternoon?

Every flavor is a solution to a user problem.

Journalism is one flavor in the information ice cream shop. You can serve Journalism in different formats — in a cone or cup, in different sizes, with nuts, or fudge sauce, or sprinkles. You can even mix Journalism with other flavors — like, say, a scoop of the Global Database of East African Women Developing Epitranscriptomic-Based Drugs, if your customer thinks they happen to go together in the moment.

Flavors exist because not everybody wants all vanilla all the time, because humans crave variety. The global culture churn means that most of my American friends under the age of 40 can say — and eat — tteokbokki; market differentiation is just good business. Offering more flavors within a single product category means you compete against yourself and build market share across various niches.

You serve an audience of one, based on contextual variables like age, phase of life, income, and specific need of the hour. It isn’t just that different users have different needs — as we know, the same user has different needs at different times. Chocolate ice cream in a glass dessert bowl with pistachio nuts and hot fudge sauce was great after dinner with friends last week, but today I need a brisk post-lunch sorbet pick-me-up with an espresso, okay?

Ben & Jerry’s talismanic cookie dough flavor came in 1984, when an anonymous fan suggested they mix cookie dough into vanilla ice cream. It’s since become an international hit.

The future — and viability — of your information ice cream shop is based on how many flavors you’re able to sell, based on how many user problems you have found to solve. And by this, I don’t mean the content-supply-driven “user-needs” models that have crept into newsrooms; I mean actual demand-based needs from actual users. Niche-based need is your friend, and growth comes from serving one niche well, then adding another, and then another.

Rather than using AI to generate more content supply to punt onto the algo-roulette table where, most often, crossed fingers pass as a business plan, organizations that use AI to find and meet existing user demand will build the viability they need to survive. Wasted content and melted ice cream are no good to anybody. Why invent demand for your weird ice cream when there’s already demand for seasonal strawberries in vanilla?

(Unless, of course, you decided to test a quick prototype and sample it with your existing customers, like these guys did.)

Actually, I’ll have a cone of your anchovy ice cream, please. No, no fudge, thank you.