Artifact §05 · Research synthesis

Seventy-six conversations, one page

Before SKADI designed anything, we talked to 76 people: freestyle skiers and snowboarders, from people learning their first 360 to a nationally ranked athlete, plus the coaches and judges who watch them. This is what that looked like from the inside, what it changed, and the two assumptions it killed.

This one is real. The numbers are from the work, the quotes are paraphrased, and nobody is named, because they gave a stranger twenty minutes and did not sign up to be on a website.

How it was run

The questions that did the work

Every interview wandered, which is the point of an interview. These four came up every time, in roughly this order, and they did most of the work.

1. Tell me about the last trick you tried to learn. How did you know whether it went well?

   Why It starts with a story, not a product. The answer to the second half is the whole problem: they did not know. They guessed, or they asked a friend with a phone.

2. What do you use today, and what does it not tell you?

   Why Shaky GoPro footage filmed by someone else. Apps that record speed and vertical and nothing about the trick. The gap between what they used and what they wanted was the requirement.

3. Who else sees your numbers, and why does that matter to you?

   Why This is the one that overturned an assumption. We expected the answer to be "my coach". For a lot of riders it was "everyone", and they wanted the numbers to be impossible to fake.

4. If this existed, what would you pay, and what would make you not buy it?

   Why The first half gave us a price. The second half gave us a list of things not to build, which is the cheaper list.

What we expected, and what was true

• We assumed

  Riders want their data as stats, and more detail is always better.

  What was true

  Younger riders wanted stats they could not edit. The point was not analysis, it was proof: to friends, to sponsors, to the group chat. A number you can change is worth nothing to them.

  What changed

  Session recaps built to be shared, and built so they cannot be altered. Social proof became a product requirement rather than a marketing idea.

• We assumed

  "Find my ski" would be the feature every backcountry rider paid for.

  What was true

  The people at the most risk trusted a five-dollar physical leash over a two-hundred-dollar tracker, and said so without hesitation.

  What changed

  The feature was cut before anyone designed it. That is the cheapest a feature ever gets killed, and it is the reason you interview before you draw.

And two things we did not expect at all.

1. The market already understood ski trackers, and was frustrated with them.

   People knew what Carv and the GPS apps did. Their complaint was specific: everything on the market measured turns and carving. None of the products people named measured the trick itself. The gap was not awareness, it was the thing being measured, which is a much better gap to find.

2. Even sponsored athletes were guessing.

   Coaches and judges said the same thing from the other side: there was no objective record of what happened in the air, so feedback was opinion, delivered from the bottom of the landing. A coach described the concept as a coaching tool before we had used the word. That sentence set the positioning.

What it changed in the product

1. Automatic trick detection became the core, not a feature.

   Airtime, rotation, height, a landing stability score, and pop timing, detected without anyone filming anything. Every product we mapped tracked the run. None of the seven measured the jump.

2. Boot-mounted, not insole.

   Two pods, strapped to the outside of any ski or snowboard boot in under thirty seconds. Better for capturing the body's motion through a trick, and it keeps clear of the insole patents the incumbents sit on.

3. The price came from the room, then got tested against strangers.

   Two hundred and fifty dollars was the average of what interviewees said they would pay, and it sat next to Carv at $249. The model priced the kit at $249.99, and the ad campaign checked whether people who had never heard of us would raise a hand at that number.

What people said, against what they did

Interviews tell you what people say. A paid funnel tells you what they do. In March 2026 we ran a small Meta campaign to a cold audience, with the price in the creative, and counted.

Thirteen is a small number, and I treated it that way. What it bought was a measured cost per signup instead of an assumed one, so the financial model started from something real: ad spend, divided by cost per signup, times a signup-to-purchase rate of 7 to 14 percent depending on the year. The sensitivity tables were built around the two inputs we controlled least (the cost per signup and that conversion rate), because those are the two numbers an investor should push on, and I wanted the answer ready.

The model also said plainly where it fell short. At the year-three run rate the venture was still about 750 units a year below break-even. I would rather show an investor that line than have them find it, and the same instinct is why the open items sit at the front of every requirement I write.

What I took from it

Nobody we spoke to talked in specifications. They described a bad day, a workaround, and a thing they wished existed, and the job was to listen for the need underneath the request, write it down precisely enough to build against, and check it against behaviour before anyone spent money on it.

Hardware raised the stakes. In software, a missed requirement is a change request. In hardware, it is a new revision of the thing itself. That made me far more careful about the gap between what people say they want and what they will actually do, and that carefulness stayed with me.

Written by Nasmaan Ibrahim for nasmaan.com. Figures from SKADI's own research and the March 2026 campaign. In August 2026 I handed SKADI's day-to-day to my co-founder.
Back to the record · Build or buy, worked through · The first fifteen minutes