One thing that's surprising about Scott Cook, the founder of Intuit, is that this intensely smart guy who's become one of the most successful entrepreneurs in history doesn't think you should listen to his ideas. He wants you to run experiments. When he was in New York recently, he told Fast Company why.
The interview below has been condensed and edited.
As you grow as a company, you get layers and hierarchy that, in theory, should make your decisions a lot better. They get reviewed, and thoughtful people look at decision. But I found myself saying, "Boy, you know, a lot of those decisions--decisions I made, decisions I participated in, decisions we took to our board--many of those decisions weren't very good."
It was perplexing, because I've got an MBA. You're kind of taught that the role of a business person is the decision making. I began to worry: Are some of the things we're doing unknowable? Is it not possible to predict consumer behavior?
You had to say that we didn't have the batting average that we imagined in our minds that we did.
I spent some time studying Toyota, because how could a loom maker--they made looms. That was their business for 50 years, 35 years--and then they decided to go into the car business after everyone else was in the car business. How could they become the world's branded carmaker, making cars that are higher quality than anyone else in the world? Making cars with parts and systems nobody could match? When Nissan sells a hybrid, it's got Toyota parts in it because Nissan can't make them. Nobody sells a hybrid luxury car that does what Toyota's done. Other guys have passive systems. Toyota has an active system.
How can this loom maker out in the Kansas of Japan become better than Mercedes and GM at the core business of making cars? I went with some professors from when I studied at Harvard, I went to Japan, to the Toyota factories, and I watched and learned together with the professors.
As one of the professors said, Toyota runs itself as a massive series of experiments from the production line worker, all the way up to the CEO, where they teach their production. Their supervisors now, they have the line supervisor manage 70 or 80 people [on] how to run experiments with their team, little scientific experiments to try out their ideas on how to make production.
They were single-variable experiments. I think that's kind of odd. Nobody else does that. We don't make autos. There were some other examples, but the one that hit home was how Google beat Yahoo. I talked to Yahoo people about it. They said, "We got beat because Google runs itself as a series of experiments run by its engineers. They are constantly trying new things at a ferocious rate. A Google chief scientist says they run 3,000 to 5,000 experiments a year. If you use Google in a week, you're likely to be in three experiments. You don't know you are, because they are experiments.
Guys at Yahoo were saying, "They just outran us. We tried management, all the stuff that management did, but we didn't have that experimentation engine.
There's a pattern here. Both companies make much better decisions because they don't rely on hierarchy, PowerPoint, persuasion. They're making decisions based on real experiments. So I said, wait a minute. Whenever reasonable, let's move from decisions by persuasion to decisions by experiment.
Three things happen. One, you make better decisions because it's actually real consumers or real production methods that aren't based on theory or a PowerPoint. It's based on real results. That's one.
Two, you enable your most junior people to test their best ideas, and when in you're doing PowerPoint presentations, whose ideas are most likely to get lost?
The third is, you get surprises more often, and surprises are a key source of innovation. You only get a surprise when you are trying something and the result is different than you expected, so the sooner you run the experiment, the sooner you are likely to find a surprise, and the surprise is the market speaking to you, telling you something you didn't know. Several of our businesses here came out of surprises.
The traditional way you do payroll is the customer comes in and they want to get a payroll system set up to pay their employees. So you come in and have to do setup first, enter all this stuff about your employees: social security number, tax status, cumulative payments to date. It takes hours to a day or two. What we noticed is when new prospects come to us, 60% of them come to us on their own payday. They don't want to wait to do paychecks if they don't have to. No the payroll company is set up this way. They all force you to do this setup, which on average takes a day.
We'd done an acquisition of a smaller startup that was about four or five years old, and one of their engineers had this idea: If the new customer wants to cut paychecks right now, we'll say "Fine, we'll cut the paychecks right now." We couldn't sell that idea inside this structure because it's so radical. The whole industry does it one way.
So he gets here, and we have an experimentation culture, so they're allowed to consider ideas like this, and they design the screens to do it, and then they did just a quick non-experiment with the consumers. They took 20 payroll customers, showed them the screen. That's not an experiment. That's just showing you the screens and asking their opinion. Experiments don't use opinion, they use real behavior. They showed 20 prospects the screen and said, "Would you decide to do payroll setup after your first round of checks," and zero percent said yes, they would use it. Zero percent. So the team stopped working on the idea.
In the meantime, we'd found (Lean Startup author-thinker-badass) Eric Ries. I discovered Eric Ries on an online video and said, "Shit, this guy's talking about the same exact thing we figured out, but he explains it so much better, and he's taken it farther that we've taken it." We immediately had him come in and start teaching it.
We have Eric now coaching in public inside the company. The team went to him. Eric looked at it and said, "Why don't you run an experiment tomorrow. Just put up the option. Show prospects the choices, and let's see what they pick. Don't build it yet. Just show them choices, and if they pick the paychecks first, setup second, we say, ‘So sorry, we haven't built it yet,' and give them a $100 gift certificate."
He challenged them. "In 24 hours, you should have this experiment up." I was in the audience watching, and in 24 hours we had the experiment up, and I remember when they showed customers and asked their opinion, zero percent said yes. When they actually didn't ask an opinion and just watched behavior, 58% clicked, "I want to cut the checks first." 58%.
They then built the functionality, but it took them a few months. Now, our payroll division is going to have their fastest customer growth in 10 years because of an experiment. Without the experimentation stuff, it never would have happened.
You've got to change your culture, because it's natural for bosses to want to decide. Now we teach our leaders that it's your job to put in the systems that enable your people to run your experiments fast and cheap and to keep making them faster and cheaper. Yield as many of your decisions off to the experiment as possible.
Large successful companies need a lean startup more than startups do. Startups have a natural discipline of the market.
[Image: Flickr user Patrick Lentz]