What Google, Genentech, and Facebook Have in Common

By Scott Kirsner |  February 9, 2017

Do innovators get fixated on designing new products and services too soon?

Luis Perez-Breva, Director of MIT’s Innovation Teams program and author of the new book “Innovating: A Doer’s Manifesto,” argues that before there’s a prototype or beta, there needs to be a hunch about a problem that needs solving. And that the first hunch is rarely on target.

    “You may be inclined to bank it all on the belief that the one solution you imagine is right. But you do not need to. Your hunch also hints at a broader space of opportunity. Most likely, there are multiple ways to have an impact in that space. The solution you imagine may be one, but for now that’s at best an assumption at risk of becoming a significant constraint.
    “…You can learn all the ways you can solve the problem by allowing yourself to be wrong about your hunch and finding out how you are wrong. The outcome can be a robust path to solving a real-world problem — no matter how wrong you are at the outset. Whether your innovating leads to the innovation you imagined, you stand to benefit the most from going about it with a certain naiveté.”

We spoke with Perez-Breva earlier this month about hunches, testing ideas, collecting feedback, and getting ready to scale.

I don’t like the buzzword ‘experimentation.’ The way it has been used the last 10 years, it leads people to think that you’re doing a random experiment here and there to try things out. In the book, I use the term ‘trial and error’ instead. I think it’s a much better description of what you’re doing. You have a hunch that you can solve a problem. You’re trying to figure out the ways you will be wrong and fail. As you continue to try things, you make the idea increasingly more robust.

The resistance I get from companies is that when they hear the word experimentation, they think of it like a random series of trials.

The initial idea is always going to be bad. It’s not going to scale. You need a continuous process of changing that idea by trial and error – taking in advice from the salesmen, and the customers, and gathering real data and evidence. The final product will look nothing like what you started with.

Author Luis Perez-Breva, Director of the Innovation Teams program at MIT.

What Google, Genentech, and Facebook have in common. It took all three of those companies essentially the same amount of time to acquire evidence that they could produce revenue. It didn’t happen overnight. It took Google about four to five years to get to the point where they could test AdWords. Genentech, it took four years to start to produce clinical trial evidence. Facebook got to advertising four years after the founding. All of those are examples of real evidence. Before you have [evidence], it’s really easy to delude yourself when you’re trying to get people to like your product. But data in a spreadsheet isn’t your user. Customer interviews are about the past. Real data is somebody spending money on something. My book’s on the Amazon best-seller list, but those are pre-sales — they aren’t real readers yet.

Nothing is new. When you look at the very beginning of every innovation story, literally nothing is new. People think they have to have this great new idea first, but innovation is mostly recombining things that exist.

People will constantly tell you why your idea sucks. That’s a universal constant — everyone has a reason in their hand about why your idea may or may not work. You need to get that reason, because it could be an insight about something that you need to figure out. Of course, not everyone has a valid reason; that’s what you as the innovator need to figure out. You want to invite feedback, especially when you’re still at a small scale of investment, when it is easy to change things.

What can you test at this scale? The thing about scaling up your ideas is that you need to ask a few questions. What might stay constant as we move up to the next scale, and what can fail and cost me a lot of money at the next scale? Is there something I can test at this scale that can prevent that failure?

The difference between startups and big companies. For startups, the cost to try something is so minimal that they don’t even think about it. The cycle to test is fast and inexpensive for them. For bigger companies, “just try it” may mean spending $1 million. Bringing that cost down is something to focus on. Big companies have layers of bureaucracy and decision-making. They also don’t realize that most ideas at the very beginning look awful. They’re not pretty. It’s a continuous process of gathering data and improving them. You need to have a set-up that lets you do partial deployments or small-scale deployments, and you need a group of people who can disperse money so that innovators in the company can focus on proving their ideas at scale.