Praise isn't a purchase.

Most founders keep a folder of compliments. The message from a friend who loves it. The demo where everyone nodded. The customer who said, "This is amazing."

It feels like proof. Jay Elliot, who spent the Macintosh years as a senior vice president at Apple, learned the expensive way that it isn't.

The darling of the USB industry

Years after Apple, Jay left his laptop behind after a flight. His email, files and settings were in that machine, and the machine was somewhere else. He asked himself a simple question: "Could I carry the work without carrying the computer?"

That question became a company called Migo. You plugged a small USB drive into a computer and your own working environment appeared: your files, your email, your settings. When you unplugged it, your work left with you.

The industry loved it. In It's the Product, Stupid, Jay writes, "For a while Migo was the darling of the USB industry." Walter Mossberg raved about it, and it won awards from PC World, Newsweek, and the Consumer Electronics Show "for design, user interface, and even the box it came in."

If you'd asked Jay then whether the market wanted Migo, he had a stack of evidence that said yes.

The review he wanted to ignore

Then PC Magazine published a review saying the software was hard to launch and get ready to use.

Jay's first instinct was to reject it. His team knew how to start the software, so from the inside it looked obvious. He ran a test he was sure he'd pass.

"I sat my teenage son down in front of the computer, pointed out the software, and said, 'Try getting this started,' knowing he would do it easily. He failed. The reviewer was right."

They rebuilt the start-up menu and the user guide. The good reviews hadn't been false, Jay says. They'd shown that the idea was useful and that an informed person could appreciate it. They hadn't shown whether an ordinary person, alone with the product, could get it working. "I had allowed proof of one part of the experience to stand in for proof of the whole experience."

His one-sentence version is the best line in the chapter: "Praise feels like the market answering, but sometimes it is only applause for the demonstration."

The watch nobody bought

Later, Migo went into a wristwatch with a USB connection. Jay loved it. "I wore it, showed it to people, and every time I met with somebody I gave them one as a gift. They all said, wow, this is great—once I explained it."

Read the end of that sentence again: once I explained it.

Watches are sold in display cases, next to other watches, and a shopper judges them by the face, the band, the brand and the price. Migo's value was the software inside, and nobody was standing at the counter to explain it. Jay puts it perfectly: "A display case is a poor software salesman: it can show you a face, a band, and a price, and nothing else."

"People liked receiving one," he writes. "The question I had not pressed hard enough was: Would they buy one? In this case, they would not."

On one of our calls, I asked Jay what he'd tell a founder about to give up. He went straight to the same problem: "people love the product but the problem is would you buy it." Then, a moment later: "They love the product if you gave it to them."

Why praise feels like proof

Look at who praised Migo: reviewers, award judges, people who got a watch as a gift, and people who'd just heard Jay explain it. Not one of them had to decide whether to pay.

That's the trap. A compliment costs nothing. A purchase costs money, time and the risk of looking foolish. So compliments come easily and purchases don't, and only one of them tells you whether you have a business.

Jay's lesson from the watch is one I'd put on every founder's wall: "If your product needs you standing beside it to reveal its value, you still have a product problem."

Separate compliments from purchases

Here's how I'd put the watch lesson to work this week:

1. Split your evidence into two piles: what people said, and what people paid for. Only the second pile counts as demand.
2. Throw out anything that needed you in the room. If they loved it after your demo, you learned that you give a good demo.
3. Ask for a real decision. A preorder, a deposit, a paid pilot, a signed order. Jay's advice to a founder about to give up is to "put the product in front of the kind of person who should buy it and ask for a real decision."
4. Test it where it's actually sold, without you. Your pricing page, your checkout, your shelf. That's your display case.

If the answer is still no after you've made the changes the evidence called for, Jay doesn't soften it. "If you have looked honestly, made the change, asked the buying question, and the answer is still no, kill this version." That doesn't mean the problem wasn't real. It means this product isn't the answer yet.

Looking back, Jay owns it plainly: "I gave too much weight to praise and too little to the buying decision." And if he did it again? "I would separate compliments from purchases."