Too early is the same as wrong

We had a good idea. We just had it about ten years too early.

In 2000 I co-founded a company called Gate121 with a few business partners. The idea was simple. Shoppers signed up and told us what they were interested in. Retailers told us who they wanted to reach. Our software matched the two, and the right people got a time-limited offer by SMS. They showed the code at the counter and got the discount.

Opt-in, targeted, straight to your phone. If that sounds like a lot of apps you've used since, that's the point.

This was 2000. No smartphones, no app stores, no maps in your pocket. For most people the internet meant a dial-up modem and a fair bit of patience. And retail marketing was still in the land of catalogues: printed, letterboxed and planned months in advance.

A little traction isn't enough

We did get some traction. A handful of major retailers ran promotions with us, and the members who used it liked it. They'd show their code, get their discount, and often buy something else while they were in the store.

But a little traction isn't a business.

Every retailer had to be talked into a new way of thinking about marketing before they could even try it. Their budgets, agencies and habits were built around print, and retail was slow to take up new technology. Then the dot-com crash hit, and anything that sounded like the internet got a lot harder to sell. On the other side, every member had to be signed up one at a time, on phones that could barely do more than make calls and send texts.

We were pushing on both sides at once, and neither side was pulling.

Early looks a lot like wrong

We weren't wrong about where things were heading. Within a decade, Groupon had built a global business on a similar idea, once smartphones made it easy for everyone.

But being right in 2010 doesn't pay the bills in 2002. Investors have a saying for this: being too early is the same as being wrong. From the outside, the market can't tell the difference between an idea that's early and one that's wrong. Either way, it doesn't buy.

That's the uncomfortable part. Being early feels like vision while you're in it. In hindsight, I think it was mostly just being early.

The problem that was already there

While we were pushing retail uphill, the conversation that changed our direction came from somewhere far less glamorous.

Talking with nurses, we kept hearing the same thing. Senior nurses were filling shifts every day by ringing around, working off phone books, handwritten lists and Post-it notes. One call at a time, until someone said yes. I wrote about where that first came up in Three months in a hospital.

The contrast was hard to miss. In retail, we were trying to create a need. In hospitals, the need was already there, every single day, and people were losing hours of their shift dealing with it the hard way. Nobody had to be convinced it was a problem. They were living it.

So we thought, hold on a sec. We'd built software to match a shopper to the right offer. Could it match a nurse's skills and availability to the right shift?

Same core idea. A very different problem, and this time it was one people already had.

That became Cascom, and later ShiftMatch. It wasn't a quick tweak. Before we built anything, we spent three months in a hospital learning how the work actually happened, and found far more underneath the problem than we expected. That's its own story, and you can read it here.

The problem shapes the technology

Looking back, the biggest lesson wasn't really about retail or healthcare. It was about which end you start from.

With Gate121, we started with the technology and went looking for a market that was ready for it. With nurses, we started with a problem people already had and fitted the technology to it. The first meant pushing. With the second, the customer was doing the pulling.

The problem shapes the technology, not the other way round. It's a lesson I kept coming back to for the next twenty-odd years.

We're doing it again with AI

Right now there's a lot of AI looking for a problem. The technology is impressive, and I use it every day. But a lot of pitches boil down to "we've got AI, who wants it?" That's close to the question we were asking retailers in 2000.

Some of those products will turn out to be right, eventually. Plenty will just be early. And from the market's side, early and wrong look exactly the same.

The ones that land will start with something people are already struggling with, and already spending time, money or patience on.

How I'd check whether I'm too early

Is anyone already solving it badly? Workarounds are the best evidence of demand you'll get. Phone books and Post-it notes told us more than any market research could have. If nobody is even trying to solve it, ask why.

How often does it hurt? For most retailers, a promotion was something they planned a few times a year. For a nurse manager, filling a shift was a problem every day, sometimes every hour.

What else has to be true? Our retail idea needed smartphones, a habit of using them to shop, and retailers comfortable with digital marketing. None of that was in place yet. List everything that has to exist for your product to work, and check whether it exists today.

How much are you asking people to change? We asked retailers to change how they marketed and shoppers to change how they shopped. For the nurses, the job stayed the same. Only the ringing around went away.

Too early, not wasted

Gate121 didn't work as a retail business. But it wasn't wasted. It gave us the matching engine, a bit of scar tissue and the lesson that shaped everything after it.

Don't get too far in front of the market. Find the problem people already have, and fit the technology to it.

If you're working on a product that feels right but isn't landing, I'd be happy to talk it through.