A few weeks ago, I caught a serendipitous Lyft with a particularly chatty driver. Normally, I keep to myself when Lyfting around town, but somehow we started talking, and I mentioned that I was headed to a conference to present my biggest marketing screw-ups. He chuckled and told me, “Smart men learn from their mistakes; wise men learn from other people’s mistakes.” A few minutes later, I got out at my stop and thanked him for giving me a great opener for my speech.

While most of the news about Zenefits focuses on our incredible growth, we — like any company — have made our share of mistakes; mistakes you’d be wise to learn from. Just take it from Mr. Lyft Driver.

Here are the top five marketing mistakes we made early on at Zenefits.

Mistake 1: We stopped doing things that didn’t scale.

Most incubators — Y Combinator in our case — encourage startups to get to market faster by building products that don’t scale (at first). No one really encourages you to do the same from a marketing perspective, but it actually holds true.

In our early days, I developed a marketing campaign that was an amalgamation of dynamic content and data. It ended up being hugely successful and contributed to a decent chunk of our first year revenue. Then, one day, Parker Conrad (Zenefits’ cofounder and CEO) came to me and said, “Great! We’re going to go from three salespeople to 10. Now, we need to scale this campaign. Let’s do every industry. Every vertical.”

I immediately told him there was no way we could scale the campaign because it would require too much data and customization. So instead, I took the easy way out and created a “scaleable” campaign by using pre-baked, non-dynamic content that would work across every industry.

Weeks later, we found out the “scaled” campaign was performing terribly. Why? Because it now had nothing that made the initial campaign awesome. Parker wasn’t too jazzed when he found out and asked me to stay as long as I needed that night to scale one industry correctly. I did as he asked, and it worked extremely well.

Two and a half years later, we have more than 100 people pulling industry data and feeding it into a custom app we built in-house, all to make this campaign work at scale.

Lesson 1: Get it to work first. Worry about scaling it later.

Do the most creative thing possible to get business. If it’s sending cupcakes with handwritten letters, then send cupcakes with handwritten letters. Then figure out how to scale it. After my experience, I firmly believe you can scale anything with enough time, money, and creativity.

The goal should always be to find something that works well and leave the worry for later; otherwise you may pass on an idea that could have completely changed your business — like I almost did.

Mistake 2: We listened to each other.

Never listen to your coworkers. It’s a terrible mistake. Let me explain.

One day, we decided to convert a few of our inbound Sales Development Representatives (SDRs) into outbound SDRs, to see if outbound sales development was a viable channel for us.

Everybody involved — the SDRs, their managers, and even my directors — were all cheering the success of the test and touting how many leads they were generating. So, naturally, we all high-fived and began hiring based on this success.

However, many months later, we dug into the data and discovered that 30 percent of the leads they were creating were actually coming from a segment of companies we didn’t really want. This messed up our hiring model in a big way.  We had planned out two years of hiring on an attainment number that was completely wrong, all because I trusted what my team was telling me. Now I only trust one thing: data.

Lesson 2: D.R.E.A.M. (Data Rules Everything Around Me)

Every conversation you have with your team should begin and end with data. Qualitative data is always helpful, but quantitative rules all. If you don’t operate in a D.R.E.A.M state of mind, you run the risk of making decisions based on how something feels — which can at times be incredibly convincing — instead of how something actually is.   

Mistake 3: We set “realistic goals.”

At the beginning of 2013, Zenefits’ first year of operation, Parker set a goal of $2 million annual recurring revenue (ARR), and we ended up falling short by a little bit. On the first workday of 2014, Parker came into the office and said, “Great! This year we’re going for $10 million in ARR” — to which I agreed because I thought we could do it. Maybe. We’re talking a very weak maybe.

The next day, Parker sat down with me and Sam Blond, our incredible VP of Sales, and said, “Okay, I changed my mind. We’re going to 20 million.” I immediately threw a fit. I went full on toddler-in-a-grocery-store. I cried. I yelled. I stamped my feet. I threw things. I couldn’t think of a single company that had done that, and I looked around at our 20-person rag tag crew wondering how we could possibly reach that goal.

Then, something interesting happened. Parker calmed me down, and asked me, “What if it was possible to hit $20 million, what would you need to do? Theoretically?” I (very grumpily) listed off all the things I would need to do, and he asked me, “Well, why don’t you do that?” Within 15 minutes, I had gone from vehemently rejecting the goal to figuring out how to hit it.

In 2014, we did $20 million. Not only did we meet our goal, but we exceeded it.

Lesson 3: Stretch Goal = The Real Goal (Ask yourself: What if?)

Whatever goal you set, people tend to come in just a little short or a little above the goal line – no matter what it is. When you create a stretch goal and get rid of the word stretch, the odds are better that you’ll achieve a greater outcome than if you set a lower, more “realistic” goal. I would have much rather fallen slightly short of $20 million than slightly beat $10 million.

When you’re put in a pressure cooker — and your back is against the wall — you do things that you didn’t think possible. If the goal feels impossible, just remember to ask yourself: “What if?” You’ll do things you can’t and won’t do when you’re comfortable and enjoying the ride.

Mistake 4: We hired people when we needed them.

Once, we were working on a giant, targeted marketing campaign. I knew there were a handful of people we would need eventually. At the time, I was hacking along on my own just fine, so I kept making it work and figured we could hire when we “really” needed help.

By the time we really needed these people, it was too late. It took 2-3 months to find the right candidates, 2-3 weeks for them to start their job, and 2-3 months to ramp. To say we were behind the 8-ball by the time they got going would be an understatement. The campaign was delayed by an entire quarter, along with all of the revenue behind it.

Lesson 4: Hire people you’re going to need 3-6 months from today.

It will take you roughly two to three months to find the right candidates, plus time to negotiate, and two weeks’ notice at their old company. And once they are finally in the seat, they still need time to set up and get up to speed.

If you’re a startup and you’re trying to show traction, your early hires will make or break your company. If you don’t have the person you need in that seat when the time comes, you’re basically toast.

Mistake 5: We focused on what worked.

At one point, we developed a marketing campaign that the whole team was enamored with it. All the SaaS metrics were perfect — we did X, and we got X demos in return. The math was simple. Do more of it, and get demos with more potential customers.

I started to put my energy — and all my team’s energy — into this one campaign. Eventually, it became so big and so successful, that it literally ate up 90 percent of my time. I didn’t care though, I was hitting goal!

Things went south really fast, however, when Parker gave me another 10x-type goal. I couldn’t 10x this specific campaign, and I had no other campaigns that could reach that goal. Four months after Parker asked me to 10x, I finally found a way to do it. Three months past the deadline.

I had become so focused on the lever that was working that I had completely stopped trying to find new levers to pull.

Lesson 5: Marketing is a rolling stone.

Our job as demand gen marketers is to find levers that work. Not one, or two, or three — dozens.  As soon as you find a lever that shows signs of being scalable, you need to put someone in charge of that lever and move on to finding the next lever. There always needs to be a rolling stone testing new levers, whether that’s the CMO, VP of Marketing, or VP of Growth.

Today, we’re growing the Zenefits marketing team so that we’ll be able to pull 20 levers at a time. Good marketing is always having 20 balls in the air, not keeping one giant ball afloat.

Matt Epstein is the first employee at Zenefits, and he currently serves as VP of Marketing. Prior to Zenefits, he was a Sr. Account Manager at Definition 6, a digital marketing agency, where he helped develop and execute online marketing strategies for Fortune 1,000 clients. He was also a D-list YouTube celebrity in his previous life — www.googlepleasehire.me.