Main idea

Iterative research doesn't do the same thing from start to finish. In preparing for Adobe Summit Paris 2026 over fifteen iterations, three phases took shape.

At the start, the iterations broaden: they serve to understand the terrain and reveal concepts, mechanisms and the first boundaries of the subject. In the middle, they reinforce and structure: better evidence, ideas pulled apart, gaps filled, relationships emerging between notes. At the end, they attack the model: the last iterations become break tests.

The question changes with the phase. It is no longer "what else can we add?" but "what could we find that would force us to revisit what we think we have understood?".

Why it matters

It gives you a steering marker: research still broadening after many iterations hasn't found its structure; research that has never tried to contradict itself isn't finished, whatever its volume.

And it separates the final figure — 82 sources in this experiment — from what really counts: the way the role of the iterations evolved.

Nuances and limits

The three-phase split comes from a single experiment within a given scope. The boundaries between phases are blurred, and phases can overlap, or even go backwards when a break iteration does find something.

Open questions

• Does a successful break iteration send the research back to the broadening phase, or only to the reinforcing one?