Greenfield product. Do you need a product strategy before you build?

A greenfield product is the easiest place to skip strategy and the most expensive place to have skipped it. There is nothing to constrain you. No legacy code, no existing customers, no capability to protect. So the first thing built tends to be the first thing thought of, and with coding agents the first thing built arrives in days.

What changed

A college student shipped BeyondWebsite, one of our own products, in three weeks, working from written acceptance criteria. The build was not the hard part. The hard part happened before the build: deciding what the product was for, who it served, and what done meant for each piece. That is a product strategy, whether you call it one or not, and for a greenfield product it is the only thing that separates a plan from a pile of prompts.

What a greenfield strategy has to say

• The first customer, singular. Not the market. The one segment and the one person inside it whose problem you are solving first.
• The one job. What they will use the product for on day one, stated so plainly that you could refuse a feature with it.
• What you are betting on. The reason someone picks this over the tool they use now. If it is "it uses AI", that is not a reason, that is the year.
• What version one will not do. Written down, because with cheap building the pressure is to do everything.

From the page to the build

With the page written, the build order writes itself. The first initiative is the job. The features under it get user stories and acceptance criteria before anyone opens an editor, and the coding agent builds from those. In one team we work with, a couple of days of written context let a junior developer build a feature in a day, tested end to end, with far fewer questions from the agent than usual. On a greenfield product the whole thing works like that, if the page exists.

And when you are wrong

You will be, somewhere on the page. That is the point of writing it. A wrong sentence on a page is fixed in a minute. A wrong sentence built into three weeks of code is a rewrite. Ship the first job, watch what people do, and rewrite the page before you rewrite the product.