There’s often so much riding on landing your first customer that it’s hard to be patient. But that’s exactly what you have to be.

While this first customer is everything to you, you’re just one of a hundred possible suppliers that the customer deals with. And buying from a startup—especially as the first customer—comes with enough risk to slow down the most aggressive company.

Our first customer was a large technology company. Our product uses data to generate better sales leads and this company was a great fit for us because it has a lot of data and a large internal sales operation. Plus it has a track record of embracing new technology.

We knew this was a do or die situation for us. If we wanted our product to achieve its potential we had to win this deal. And we did—15 months after we first started talking.

Looking back, there are several things that helped us:

Don’t be pushy. As much as we wanted an answer, we made sure never to put the customer in a situation where they had to give us a yes-or-no answer. If we had forced a decision at any point earlier, the answer almost certainly would have been a no.

Listen more than you talk. This was a big lesson for us. As a founder, you have a vision and you naturally want to tell people what you can do for them. You need to avoid this temptation. Every time we met with the customer, we made a real effort to set aside our agenda and listen to what they wanted us to do for them. We were then able to match that up to what we were capable of. They kept saying that we really understood them.

Do something different. If we had just made the same Lattice Engines pitch over and over again followed by a demo we never would have made the sell. We set ourselves a goal of providing something of value at every meeting and held ourselves to it. Initially, we asked for some of their data and later showed them some patterns we discovered. Then we did it again with a larger data set and made some recommendations based on our findings. We did this sort of thing again and again. Also, asking for some data—a hard request to say no to—gave them a question to answer other than do they want to buy our product.

Be flexible. After nine months, the customer said it wanted to do a real experiment to see if our software actually worked. We agreed in part because we were confident that it was going to work, but also because it was the only thing we could do. They knew they would be our first customer. They had all the leverage.

For the pilot, the customer set up a control group and said that if the group using our software performed 3% better over six months they would buy it. We got 11%. We had to restructure the contract because the way we had written it initially made that much improvement too expensive.