The difference between nice reading and a report worth paying for

08 September 2026

The difference between a report that reads nicely and one worth paying for is usually the method — the part that says how you know what you claim. Most people are never handed that expectation. I wasn’t either.

I saw the gap most clearly in a piece of desk research I’d set a group of students. I asked them to write up how they did it — a short methods section. Most left it out.

My first thought was the usual supervisor one, and I won’t even spell it out because it was wrong. What we worked out together, though, seemed to be that they were never asked to consider that a “Methods Section” could also be needed for a writing assignment. Materials and Methods is something you learn in a lab. Nobody had ever asked them to apply it to reading, searching, and thinking, too. (Which, by the way, is something that, as far as I can recall, I was never asked or taught to do at uni either.) In a commercial report, like a market intelligence report, sources, method, assumptions and reasoning are always explained as a matter of course; it is the difference between nice reading and a meaningful report worth a lot of money. I concluded that students rarely get handed such an expectation for the work they’re doing. The gap I saw was probably in the teaching, not in them.

At the same time I was watching a second pattern across the projects I run. Students were using AI to produce work with no method behind it — no direction, no record of what they had asked or checked, no decision they could point to as their own. They weren’t cheating; they followed my instructions to use AI. But I’d asked them to use a powerful tool, and they had no framework that made deliberate use the obvious default.

There was material to work from. I had a draft AI guide of my own, written for one project, sound in its instincts but shaped like an internal brief. And I had a photograph of a university faculty’s AI poster — the familiar four zones, from misuse to encouraged. The poster was useful, but it was pointed the wrong way for me: it told students which zones to stay out of. I wanted the opposite emphasis. Use it — but with intention, method, and a record of what you did.

A few ideas did most of the shaping.

The first was to treat AI as a way to create a team, not simply a tool you ask questions. You are the captain of your ship, the project leader. The models are specialist team members you assign, direct, and answer for, and — most importantly — you do not hand any of them a decision to make. That one move changes the posture from “asking the machine” to “running the project.”

The second came from years in drug development, where you never assemble a team before you know what stage the project is at — discovery asks for different people than late trials do. The same goes for a virtual team. Work out where the project sits before you decide who, or what, you need on it.

The third is the one I care about most, because it is the one the tools erode. AI knows how things are described. It knows NOTHING about what it costs to fail at something, what it takes to recover, or what it means to have made a decision that mattered. It does not care about the consequences of its actions. And no conscience or ethical concerns to trouble it, either. So the method in the guide builds in a deliberate step to bring real people into the loop — peers, teachers, someone older. A person’s experience is not made obsolete because the tooling changed.

The guide was built the way it tells students to work: AI-assisted, directed at every turn, with a record kept of how it was done. It carries an appendix documenting its own making — the same two-part method it asks of the reader, turned on itself. That is not cleverness for its own sake. It is the only honest way to publish a guide about method.

It is a Version 1 (even if it’s V1.7). The best version will be written by the people who use it, mark it up, and tell me where it is wrong.

This post is part of Data on Deck - Picture yourself in Enterprise’s captain’s chair with a bridge full of Datas. (Or: How to Use AI in Your Project.)