IKEA, building with AI, and UX

Hi!

After shutting down Sputnik Legal, I’ve started building Sputnik Intelligence (yes I’m lazy at naming things), and while I’m building that I’m realizing a bunch of things about UX, design and AI, so I’m going to send out the occasional Wednesday newsletter that is less focused on AI/UX news (the Friday newsletter keeps doing that), and more on the insights from building. I hope it’s interesting!

So I’m finding that building with AI makes it much easier to focus on the few key things that customers really want. I’m still figuring out what that is, but Sputnik Intelligence is about getting data about what actual humans care about (that’s why it focuses on podcasts), versus the slop the internet is filling up with.

For IKEA, they have their brand pretty clear, what customers want from them is great design at an affordable price. They call it design-to-cost. The price tag comes first and design works backward from that number.

A great story is the TROFÉ mug in 2005 (Bloomberg). IKEA handed the team a price before anyone drew a mug: 5 Swedish kronor, about 50 cents, what the product developer called “a real knockout price.” Design started from that number, with the factory in the conversation.

They redesigned the mug 3 times, and every redesign was a logistics decision. The first version fit 864 mugs on a pallet. A new rim, borrowed from flowerpot designs, got that to 1,280. A shorter mug with a reworked handle got it to 2,024. The shorter mug also took less space in the kiln at the Romanian factory that fired it. Shipping costs dropped 60%. The price stayed at 50 cents.

This runs through the whole company. Marcus Engman, IKEA’s head of design at the time, told Dezeen in 2015 that designers, engineers and product developers work as one team from day one. “One centimetre there could maybe mean 10 euros in the end on each and every product.”

It hasn’t changed. Johan Ejdemo, who runs IKEA design today, told Fast Company this spring that his team works by “distributing value and cost within the product.” PAX wardrobe shelves get extra material density where the drill holes go, and less everywhere else.

Amazon runs on the same logic with different levers. At AWS re:Invent in 2012, Jeff Bezos said retail customers want cheap prices and fast delivery (TechCrunch). Add selection and you have the 3 things Amazon has optimized for since 1994. None of that has changed in 30 years, and none of it will.

I’m finding AI very handy to do this: bring all the different disciplines together from day 1, instead of the typical: someone specs the ideal experience, then hands it to engineering to figure out what’s feasible and what it costs. Or the reverse: build what the tech allows, then find a story for the customer. Either way, price and possibility get decided in different meetings, by different people.

Decide what won’t change in your market. Selection, price, speed, whatever the durable thing is. Then have the tech conversation alongside it instead of after it.

For Sputnik, I know one of the durable things already: people want signal from real people, not machine-written slop, and they want it fresh. I haven’t landed on the other two yet.

More on this as I keep building.

Health and happiness,
Peter
PS: So Sputnik is really powerful for alerts about what real humans are saying in podcasts and newsletters about your brand, your product, your CEO, your competition, you name it. Real-time alerts, trends, what is the competition talking about, etc. If that’s interesting to your or your company, just reply, we can discuss.