In my last article, I looked at businesses such as Pets.com, Webvan and Quibi. They all had something in common. There were genuine market signals supporting their ideas, but those signals were interpreted incorrectly.

There is an obvious counterpoint to that argument: sometimes an idea looks like a bad one precisely because the most obvious market evidence points in the wrong direction. Customers say they don't need it. Existing products seem perfectly adequate. Competitors have already tried something similar. The proposed price looks ridiculous. Or the whole thing simply sounds too strange to work.

And yet some businesses get it right. Not because they ignored the market, but because they understood the signals better than everyone else.

Here are five examples.

1. The iPhone: nobody was asking for a $500 phone without a keyboard

It's difficult to remember quite how strange the original iPhone proposition looked in 2007.

The mobile phone market was already mature. Nokia dominated globally. BlackBerry had built an enormously successful business around smartphones with physical keyboards. Microsoft, Palm and others were already in the market.

Then Apple introduced a phone costing $499 or $599 in the US, tied to a two-year contract, with no physical keyboard.

There were plenty of reasons to be sceptical. Microsoft CEO Steve Ballmer famously laughed at the price and questioned its appeal to business customers. BlackBerry executives initially struggled to believe Apple had produced what it claimed.

And consumers weren't exactly demanding an iPhone. Very few people were saying: "What I really want is a large piece of glass with a computer operating system inside it."

But that wasn't the important market signal.

Consumers were already carrying several devices. They used phones for calls and messages, iPods for music and increasingly used computers to access an expanding internet.

Apple saw that these behaviours were converging.

It also recognised something more subtle: existing smartphones were designed around the limitations of phones. Apple approached the problem from the opposite direction and effectively shrank a computer into something you could carry around.

The opportunity wasn't revealed by asking consumers what features they wanted on their next Nokia. It came from understanding how their behaviour was changing.

That's a very different kind of market intelligence.

2. Airbnb: strangers aren't supposed to sleep in your house

Imagine pitching Airbnb before Airbnb existed.

The proposition sounds terrible: People are going to allow complete strangers into their homes. Travellers will sleep in those strangers' spare rooms. They'll arrange it through a website run by a startup they've never heard of. And somehow both sides will trust each other enough to exchange money.

There were perfectly rational reasons to believe consumers wouldn't do it. However there were signals suggesting something interesting underneath.

Hotels were expensive, particularly during major events. Travellers already looked for cheaper alternatives. People had spare rooms and wanted additional income. Craigslist demonstrated that strangers were willing to transact directly with one another online.

The founders themselves famously rented out air mattresses in their apartment when hotels in San Francisco were full.

The important signal wasn't: "Consumers want to stay in strangers' houses." Almost nobody would have described their need that way.

The signal was: "People need somewhere affordable to stay, existing supply sometimes fails them, and some consumers are willing to trade familiarity for price, location or experience."

Airbnb connected those signals. It then built mechanisms such as profiles, reviews, payments and reputation systems to reduce the obvious trust problem.

Sometimes the market opportunity isn't a demand for the product you're proposing. It's an existing behaviour waiting for someone to remove the friction around it.

3. Spotify: why would anyone pay for music they could get for free?

By the mid-2000s, the music industry's problem seemed obvious. Consumers had discovered that digital music could be copied essentially for free. Napster had demonstrated enormous demand for online music, but it had also demonstrated consumers' willingness to pirate it.

That could easily have led to a depressing conclusion: People aren't willing to pay for digital music.

Spotify interpreted the signal differently.

Perhaps people weren't fundamentally opposed to paying. Perhaps piracy was simply offering a better consumer experience.

Downloading music illegally was often easier than buying it legally. Consumers wanted immediate access to enormous music libraries without purchasing individual albums or tracks.

Spotify's proposition was therefore unusual. Instead of trying to make piracy impossible, make paying more convenient than piracy. Give people instant access to almost everything. Make discovery easy. Synchronise it across devices. And offer a free version for people unwilling to subscribe.

Spotify wasn't fighting the behaviour the market had demonstrated.

It built around it.

That's an important distinction. A negative market signal doesn't always mean there is no opportunity. Sometimes it means the existing business model is wrong.

4. Netflix: the opportunity wasn't really DVDs

Netflix began with another proposition that could easily have been dismissed.

Blockbuster already existed. Its stores were everywhere. Consumers could walk into one, choose a film and take it home immediately. Why would they order DVDs through the post and wait for them to arrive?

But Netflix recognised friction that the dominant market model had normalised. People didn't particularly enjoy making two trips to a video rental store. They didn't like late fees. They wanted a larger selection.

The interesting part of Netflix's history is that the company continued reading those signals as the market changed. DVD-by-mail wasn't the ultimate opportunity. Convenient access to entertainment was.

As broadband improved, streaming became increasingly practical. Netflix was prepared to undermine the very DVD business that had made it successful because the underlying customer behaviour mattered more than the existing product.

That's something established companies routinely struggle with. They monitor demand for the thing they currently sell rather than the reason customers buy it.

Blockbuster was in the video rental business. Netflix increasingly understood that it was in the convenient entertainment business.

That distinction turned out to matter enormously.

5. Amazon Web Services: why would companies rent someone else's computers?

AWS might be the most interesting example of all.

For decades, serious businesses bought their own computing infrastructure. Servers were assets. Companies bought them, installed them, maintained them and employed people to operate them. The idea that important corporate systems would instead run on computing infrastructure rented from an online bookseller would have sounded bizarre.

Yet there were signals underneath the existing behaviour.

Businesses didn't actually want servers. They wanted computing capacity.

Buying servers meant forecasting future demand, spending capital upfront, maintaining excess capacity and dealing with considerable technical complexity.

Amazon had encountered many of these problems itself while building the infrastructure required to support its enormous retail operation. AWS turned infrastructure into a service.

Companies could consume computing resources when they needed them and pay accordingly. Again, customers weren't necessarily asking for the product.

The market signal was the friction surrounding the existing solution. That distinction matters.

Customers don't always know what the solution looks like

These examples create an interesting problem for traditional market research.

If you'd asked consumers in 2006 what they wanted from their next mobile phone, how many would have described the iPhone?

If you'd asked travellers whether they wanted to sleep in strangers' spare bedrooms, what would they have said?

If you'd asked companies whether they wanted their infrastructure hosted by Amazon, would the response have predicted AWS?

Probably not.

But that doesn't mean customer evidence is useless. It means you need to ask the right questions.

Instead of asking: "Would you buy this?"

Ask:

"What are you doing now?"

"What frustrates you about it?"

"What alternatives have you tried?"

"What are you already spending money on?"

"Where are people creating their own workarounds?"

"What behaviour is changing?"

Those questions investigate the market without requiring customers to invent the solution for you. And that's important because customers are usually much better at describing their problems than designing businesses to solve them.

The signal isn't always where you expect it

There is another connection between these businesses and the failures I looked at previously.

Pets.com saw that people spent enormous amounts on pets.

Correct.

Quibi saw that people increasingly watched short-form video.

Correct.

Webvan saw that people would eventually buy groceries online.

Also correct.

The problem wasn't a lack of market signals. It was the conclusions drawn from them.

The successful businesses did something different.

Airbnb didn't conclude that people were desperately searching for strangers' bedrooms. It recognised friction in accommodation and unused supply.

Spotify didn't conclude that piracy proved consumers wouldn't pay for music. It recognised that piracy demonstrated enormous demand for convenient digital access.

Netflix didn't conclude that people wanted DVDs delivered through their letterboxes. It recognised that people wanted easier access to entertainment.

AWS didn't conclude that businesses wanted virtual servers. It recognised that businesses needed computing capacity without necessarily wanting the cost and complexity of owning the infrastructure.

And Apple didn't simply ask consumers what features they wanted added to their existing phones.

It looked at what people were increasingly doing with digital devices.

The difference is interpretation.

This is why "idea validation" makes me uncomfortable

This is also why I deliberately don't describe SignalCraft as an idea validator. Validation implies that you're trying to establish whether an idea is right.

That's dangerous.

If you begin research with the objective of validating something, confirmation bias is almost built into the process. There is almost always supporting evidence somewhere.

A huge market.

A favourable growth forecast.

Some unhappy customers.

A relevant trend.

A Reddit post asking for something similar.

Put enough of those together and practically any startup can start looking promising.

But Airbnb also demonstrates the opposite problem.

If your research simply asks whether people currently buy accommodation from strangers, you could have rejected one of the most successful marketplace businesses ever created.

Good market intelligence needs to do both.

It needs to find the evidence supporting an opportunity and the evidence challenging it.

It needs to distinguish what customers say from what they actually do.

It needs to identify existing alternatives, workarounds and behaviours.

And it needs to be very careful about the conclusions drawn from those signals.

That's what I've tried to build into SignalCraft. It doesn't answer:

"Is this a good idea?"

It tries to answer:

"What does the available market evidence actually support?"

There's an important difference.

The objective isn't to predict the next iPhone

No market research process could have told Steve Jobs in 2007 that the iPhone would become one of the most successful products ever created.

That's not what market intelligence is for. There will always be uncertainty. Entrepreneurship involves making decisions without knowing the outcome. The objective is to make those decisions with a better understanding of what is already happening around you.

Where is customer behaviour changing?

Where are people dissatisfied?

What workarounds already exist?

What are people paying for?

Which assumptions have strong evidence behind them?

Which assumptions are mostly speculation?

And what evidence would make you reconsider the opportunity?

Pets.com and the iPhone sit at opposite ends of a useful lesson. A huge existing market doesn't necessarily mean you've found an opportunity. And a market that doesn't obviously exist yet doesn't necessarily mean you haven't.

The difficult part isn't finding signals.

It's understanding what they mean.