Prototype-led positioning

Every positioning exercise in the twenty years before 2024 produced the same artifact: a deck. Fifteen slides, maybe twenty, laying out the category, the competitive frame, the messaging hierarchy, the go-to-market. The deck was the deliverable because the deck was the only way to align a team on an abstraction before building the thing.

In 2026, that logic has inverted. The inversion is specifically enabled by the 2024–2026 generation of AI-assisted prototyping tools — Claude Code, Cursor, Figma Make, v0. Before these tools, the cost of a working prototype exceeded the cost of a deck by roughly 100x. After them, the two costs have converged. A working prototype is now faster to produce than a positioning deck, more legible to the team that has to build against it, and produces better market feedback than any amount of messaging testing.

This essay makes the case for prototype-led positioning — the practice of treating a working, shippable prototype as the primary positioning artifact, with the deck (if it exists at all) as a derivative summary.

The deck era and its assumptions

The deck era rested on three assumptions, each of which is now false.

First, that a team could not align on an abstraction by building it — the build was too expensive, so alignment had to happen on slides. The canonical positioning deck, codified in April Dunford's Obviously Awesome (2019), assumed a gap of weeks or months between a positioning decision and a shippable artifact that tested it. That gap closed in 2024.

Second, that customers and stakeholders could usefully evaluate positioning from a deck. This was always a fiction — the feedback loop between a positioning deck and a purchase decision was weeks long and bi-directionally noisy. Decks produced consensus inside the building, not outside it.

The deck era rested on three assumptions, each of which is now false.

The author, 2026

Third, that decks compounded. That is: the investment in a positioning deck paid back in reusable messaging, sales enablement, and internal clarity. In practice, most decks were rewritten every six months, their intellectual core lost to turnover. The compounding was illusory.

What replaces the deck

The prototype. Specifically: a prototype that (a) runs in a browser, (b) is interactive to the point of being usable, (c) is built in under a week by the same person who would have written the deck. Not a Figma mockup — a Figma mockup is a deck with a different aspect ratio. A shippable, working prototype that a stranger can use without a voiceover.

The working prototype does three things the deck cannot.

It tests the thesis against the physics of the medium. A positioning statement that cannot survive contact with an implemented interface is not a positioning statement — it is a wish. The 1847 patent office did not grant patents on sketches of machines; it granted them on working models. The discipline is not new.

It legibly compounds. A prototype, once built, is a git repo. It forks. It is extended. Its primitives — components, flows, data models — are reusable in a way that a slide deck has never been.

It produces feedback from the market in hours, not months. You can show it to a customer. You can put it behind a sign-in wall. You can publish a URL and watch what happens.

The practice

Prototype-led positioning as a practice has three disciplines:

1. Every positioning claim must ship as a working artifact before it is allowed into a deck.
2. The artifact, not the deck, is the object of critique in the positioning review.
3. The positioning deck, if produced at all, is written from the prototype — a derivative summary for people who cannot use the prototype themselves.

This is how the rest of this site is structured.