Why the Best Products Are Deliberately Incomplete

There's a famous story about Betty Crocker cake mix that gets told a lot in product circles. In the 1950s, General Mills launched a mix that only needed water. You stirred, you baked, you had cake. It was the most convenient baking product anyone had ever made, and it flopped.

They hired a psychologist named Ernest Dichter to figure out why. His diagnosis: the mix was too easy. Housewives felt guilty using it. They felt like they weren't really baking. So General Mills took the powdered eggs out of the mix and told customers to crack in a fresh egg themselves. They deliberately made the product harder to use. And sales took off.

This story gets told in every product design class and every behavioral economics talk. Usually the lesson is something about the IKEA Effect, or about adding friction to create engagement. But I think most people miss what's actually interesting about it.

The story everyone gets wrong

For one thing, the standard version isn't quite true. The historian Laura Shapiro spent years researching this for her book Something from the Oven, and the reality is messier. Fresh eggs genuinely produced a better cake. Powdered eggs in the 1950s still tasted faintly of the chemical rations soldiers had complained about during the war. So some of the sales improvement was just, you know, a tastier product.

There's also the Pillsbury problem. While Betty Crocker went all-in on fresh eggs, Pillsbury kept selling complete mixes with powdered eggs. If the egg theory explained everything, Pillsbury should have been crushed. They weren't. They're still a major brand.

What Shapiro found was that the real turning point for boxed cake wasn't the egg at all. It was frosting. Both companies started running ads that showed housewives making elaborate cake decorations. Birthday messages, colored icing, layered designs. The whole marketing approach shifted from "this saves you time" to "look what you can create." And that worked.

But here's what I think is interesting. Whether you believe the clean version or the messy version, you end up at the same insight. The egg, the frosting, the decorating. They're all doing the same thing. They're giving the user a role in the outcome. The cake mix became popular not when it became better, but when it became incomplete. When it left a gap for the person to fill.

I keep noticing this pattern in startups. I've been building at an early-stage company for the last few years, and the question of how much to automate and how much to leave for the user comes up in almost every product decision we make.

The most successful products tend to be the ones that ask something of their users. Not a lot. But something. Enough that the user walks away feeling like the result is partly theirs. And the products that do everything for you, the ones that seem like they should win on pure convenience, often don't inspire the loyalty you'd expect.

The IKEA Effect

In 2011, a group of researchers at Harvard, Yale, and Duke actually measured this. Michael Norton, Daniel Mochon, and Dan Ariely gave people unassembled IKEA storage boxes and asked them to build them. Then they compared what the builders would pay versus what a control group (who got identical pre-assembled boxes) would pay.

The builders offered 63% more. Same box. The only difference was that they'd put it together themselves.

They ran a similar experiment with origami, and it got even stranger. People who folded their own cranes, badly, valued their lopsided creations at the same price as origami made by experts. Not roughly the same. Statistically indistinguishable. They genuinely believed their crumpled paper was worth what a master folder's was worth.

The researchers called this the IKEA Effect. And the most important finding wasn't the main result but the boundary condition: when participants assembled something and then had to take it apart, or when they failed to complete the assembly, the effect vanished entirely. It wasn't effort alone that created the attachment. It was effort that ended in something you could point to and say, I made that.

This matters because it tells you what kind of difficulty to design for. The Egg Theory (which is what I've started calling this whole cluster of ideas) isn't "make your product harder to use." That's a misreading. It's more like: leave a specific kind of gap. One that the user can close. One that, once closed, makes them feel ownership.

The blankness that almost killed Notion

Notion is probably the purest example of this I've seen in software.

When Notion first launched, it was essentially an empty page. No templates, no pre-built structure, just a blinking cursor. For most people this was bewildering. The product could theoretically become anything, but that meant in practice it became nothing. People opened it, didn't know what to do, and left. Growth was painfully slow.

But something interesting happened with the users who stuck around. They started building. Not just taking notes or making to-do lists, but constructing elaborate personal systems. Dashboards, project trackers, life wikis, reading databases. And then they started sharing what they'd built. A whole template economy emerged. Some creators made over a million dollars selling Notion setups. The company eventually reached 100 million users.

The thing to notice is that Notion's features aren't particularly unique. Lots of tools have databases and kanban boards and wikis. What's unique is that every Notion workspace is something its user built from scratch. And that changes the economics of switching completely. Moving to a competitor doesn't just mean learning new software. It means abandoning something you constructed with your own effort. Which, as the IKEA Effect research tells us, you're going to overvalue by about 63%.

So the blankness that almost killed Notion turned out to be its moat. This is a pattern I think is underappreciated: the thing that makes a product hard to adopt is often the same thing that makes it hard to leave. The difficulty is the feature. You just have to survive the initial period where it looks like a bug.

You see the same dynamic everywhere once you start looking. Build-a-Bear charges parents $25 to $50 for a stuffed animal they could buy pre-made for a fraction of the cost. But the kid picks the skin, stuffs it, stitches a heart inside, names it, makes a birth certificate. The company has done over $2 billion in cumulative sales since its founding. That's not because the bears are high quality. It's because the ceremony of making it transforms a commodity toy into something the child (and therefore the parent) considers irreplaceable.

Substack works this way too. It gives writers a blank page and a subscriber list and asks them to build everything else. The name, the design, the voice, the cadence. Every issue is an act of construction. Writers stay on Substack even when competitors offer better economics, because they're not just using a tool. They've built something on top of it. And you can't easily port the feeling of ownership to a new platform.

Figma gives designers a canvas, not a finished layout. Minecraft gives players blocks, not a world. Lego gives kids plastic, not a toy. In every case, the product is deliberately less than complete. And in every case, the incompleteness is what creates the attachment.

The egg was never for the cake

There is a darker version of this insight, though, and I think it's worth being honest about.

The IKEA Effect doesn't just make people loyal to good products. It makes people loyal to bad ones. Norton and Ariely's origami experiment showed this clearly. Participants rated their own amateurish creations as equivalent to an expert's. The labor didn't improve the product. It impaired the person's ability to judge it.

This is the same mechanism that keeps founders working on bad ideas for years. You build something, you pour effort into it, and the effort itself becomes evidence (in your own mind) that it must be worth something. It's the sunk cost fallacy, except it doesn't feel like a fallacy. It feels like love. It feels like pride. That's what makes it dangerous. The emotion is real even when the valuation isn't.

I think this is also what's happening with a lot of onboarding flows in software. When an app asks you to "customize your experience" in the first five minutes, it's not because your preferences are meaningful on day one. They aren't. It's because ten minutes of building inside a product makes you dramatically less likely to leave. The customization step is an egg. It exists to bind you to the product, not to improve it.

The egg was never for the cake. The egg was for the baker.

What AI gets wrong about convenience

And I think this connects to something broader about what people actually want from products and tools. Betty Crocker's engineers thought they were selling convenience. They had every reason to think that. The whole arc of consumer technology is about removing effort. But convenience turned out to be the wrong frame. The housewives of the 1950s didn't want a cake that made itself. They wanted to feel like bakers.

The output was secondary. The identity was primary.

I think this is still true, and if anything it's becoming more true as AI makes it possible to automate more and more of the work people used to do. The question isn't just whether a tool can do the job. It's whether it leaves enough room for the user to feel like they contributed. Take that away and you might have a more efficient product. But you'll have a less loved one.

The best products are not the ones that do the most. They're the ones that do just enough less. Enough that the person using them walks away feeling like the result was theirs. That's the Egg Theory. General Mills stumbled onto it in 1952 by accident, and seventy years later, the most successful platforms in the world are still running the same play on purpose.