Method

Our method has four steps. We follow them in order, except when the order is the assumption we are testing.

1. Map the surface

We interview the people closest to the idea and write down everything it depends on. Not the goals, the dependencies: the facts that have to stay true for the plan to work.

2. Find the curvature

Some parts of a plan are flat: change one thing and the rest stays put. Other parts are curved: a small change in one place bends everything nearby. We find the curved regions, because that is where plans fail.

3. Compute the normal

At the point of highest curvature, we ask what direction is perpendicular to every option currently on the table. This is the question the method exists to answer. It usually takes an afternoon. Occasionally it takes much longer, and we have learned not to rush it.

4. Project and test

We move the idea along the normal and see what happens. Sometimes it improves. Sometimes it collapses, which is also useful information, delivered earlier and more cheaply than the market would have delivered it.

A note on step three: in rare cases an idea turns out to have no surface at all, and therefore no normal. We do not bill for those engagements. We do write them up.