Finding a prompt topic worth selling

A nonprofit treasurer wants to sell a prompt, and she's torn between three ideas: "professional emails," "a productivity assistant," and "the annual meeting report of a small nonprofit." The first two are within anyone's reach and already exist everywhere. The third is the one she does every year, with its minutes, its votes, its quorums to check, and its financial statement to present to volunteers who are worn out by numbers. It's also the only one that another treasurer might want to buy.

Choosing your topic is the step that decides what follows. A well-made prompt on a topic nobody is looking for stays invisible, whereas a decent prompt on a specific need finds its reader more easily. Here is a method for sorting, before you spend evenings on it.

Start with what you already do

The most reliable starting point is your own work. For a week, note every time you ask an AI for something, or could have. Also reread your conversation history: the requests you touch up or follow up on several times are candidates.

You can ask the AI to help you sort through what you tell it:

I am [YOUR JOB OR ROLE]. Here is the list of tasks I do every month:

[LOOSE LIST]

Ask me one question at a time to identify, among these tasks, the ones that are repetitive, that call for professional knowledge, and whose result can be judged on the spot. When you have enough to go on, suggest three candidate tasks to turn into a prompt, with, for each, what the user would have to supply and what they would get.

In its answers, hold on to the tasks where you have things to say that your reader doesn't know: a frequent mistake, a rule of your industry, a special case that wastes time.

Listen to what others ask you

Your professional circle tells you, without meaning to, what's missing. What questions do beginners in your field ask you, over and over? What document template are you asked to share? Which step costs your colleagues an evening? Each of these situations is a possible topic, with a valuable quality: you already know the person who would need it.

Look at what already exists, too. Browse the catalog of the marketplace where you plan to publish: the topics already covered, how they're described, what remains empty. A topic that's already well covered requires you to bring something truly different, like another audience, another method, or a better result.

A grid for sorting your candidates

Score each idea from 0 to 2 on these six questions, without telling yourself stories:

It's not a scientific calculation, just a way to force yourself to answer honestly. An idea that scores 0 on knowledge or on gap probably isn't a product: you can keep it for yourself as a simple work tool. An idea with no 2 at all deserves to be narrowed down. The articles on selling prompts and on the listing that sells develop the buyer's point of view.

Narrow the topic

A topic that's too broad describes nothing. Three ways to shrink it:

1. By audience. "Meeting minutes" becomes "annual meeting minutes for a small nonprofit."
2. By task. "Helping a freelancer manage clients" becomes "chasing an unpaid invoice without offending the client."
3. By output format. "Content for social media" becomes "a week of posts for a restaurant, based on the market delivery."

The best sign of a well-narrowed topic: you can write its listing in one sentence, with the audience, the task, and the result. If you can't, it's still fuzzy.

What no method guarantees

A well-chosen topic improves your odds without promising sales. We have no reliable figures on which topics sell best, and the lists of "profitable niches" that circulate generally rest on nothing verifiable. A niche that seems obvious may also interest no one.

A low-cost test is to ask three people in the trade: "if a prompt did this, would you use it?" Listen to the hesitations, and ask what's missing before imagining the price they'd pay. Their reaction to your explanation of the task matters more than a compliment on the idea. If you then want to set the price, the article on the price of a prompt gives a method.

This week

Open your AI history or last month's calendar and write down five tasks you repeated. Apply the grid to each, keep the best two, then put the question to three colleagues or acquaintances in the trade. The one that draws real demand, for example "can you send it to me?", is your next prompt. Prepare it with the method for reusable templates.