Building a shock-absorbing insole from one video
A founder in California was starting an insole brand — NUVE — and sent us a video with one question: could we find an insole that behaved like that?
01 — Reading the brief
The video showed three insoles, and the founder had already taken the brief apart for us. A green insole for the shock absorption they were after. A Nike for the shape. An ordinary retail insole for the baseline — what a normal insole does, so the target had something to beat.
So this was never “copy this.” It was three requirements carried by three different objects. What was missing was the translation: “absorbs like this one” is not something a factory can quote against. It had to become a material, a structure, and a test a sample could pass or fail — otherwise every supplier would answer a slightly different question.
02 — Writing the spec before contacting anyone
We wrote the product definition first, and it ran to five pages. Two size runs, 34–38 and 39–44, with a stated length for every size. A width ceiling on each run — 8.5cm and 9.2cm — so the insole would still fit a smaller shoe. Carve-mold zones where it needed to flex, a grip texture underneath, and arch support.
The material was named rather than gestured at: Sorbothane, or a Sorbothane base under a memory-foam top if no single material could both kill the impact and stay comfortable. The lining was a decision with a reason attached — basic polyester was the cheap answer, a breathable mesh cut sweat, bacteria and smell, so mesh was the target.
The spec kept the founder’s references apart rather than blending them into one vague target. Shape was assigned to the Nike SB Solarsoft, in as many words: match it “without molds, branding or any mark.” A benchmark is how a subjective requirement becomes measurable. Copying is a different thing, and page four ruled it out in writing. The finished insole carries NUVE’s logo and nobody else’s.
The absorption target was not a picture at all. The founder shipped us the green insole, so the behaviour we had to match sat on the bench for the whole project.
That changed the question we put to suppliers from “can you make an insole” to “can you hit this” — and it gave all of them the same target instead of their own interpretation of a video.
03 — Sampling until something passed
Then we sampled, tested, and eliminated. It took seven rounds. Six of them failed, and each failure removed a class of material from the list — which is the only thing a failed sample is good for. Because the founder’s insole was on the bench, every round was measured against the real target instead of against our own spec sheet: the spec was our translation of the brief, and the insole was the brief.
The result
The last test was not run against our spec sheet. It was run against the founder’s own insole, side by side — same bottle, same height, one drop each. Our sample matched it: breathable mesh over a shock-absorbing base, in NUVE’s own shape.
What the founder ended up with was not a list of factories. It was a brief broken into parts, a spec any factory can quote against, and a material combination proven against the object they sent us rather than by claim.