A universal challenge with SaaS businesses is breaking through plateaus in growth. At heart, this is just basic math. Churn is a percentage, and growth is usually linear. Customers don’t always churn only because your product isn’t providing value. Often they go out of business, change direction, or are acquired and forced to adopt new software. Simply put, the more customers you have, the more customers you have to lose.

On the other hand, barring viral growth–which is rarer than people would have you believe–growth is generally linear. As you acquire customers linearly, you will reach a point where churn intersects growth, and you’ll plateau.

Let’s look at an example. Imagine you have 100 customers, and you’re acquiring 10 per month. Now, since it’s early, imagine you have a high monthly churn of around 5%; that is, 5% of your customers cancel each month. With 100 customers, you lose 5 each month. But since you’re gaining 10, you’re still growing by 5. When you eventually hit 200 customers with 5% churn, you’ll lose 10 customers per month. If you’re still only gaining 10 per month at that point, then you will have plateaued.

In such cases, you either need to reduce churn, or increase the number of new accounts–or both, ideally. The bad news is you’re likely to have regular plateaus. This is completely normal, and it can be exhausting when you can’t find the thing that helps you break through. The key is to remember that you can break through, but only by trying new things.

The good news is that since churn and growth are formulaic, you can see plateaus coming a mile away. If you can predict when they’re going to happen, you can also start taking steps to avoid them.

Reducing churn is your best bet. Pouring more water into a leaky bucket is a never-ending battle. Better to plug the holes and then get back to work filling the bucket. After reducing churn, invest in improving the experience of new customers. The upside of these two tactics is that they’re effectively free. These people are already interested in your product. You didn’t have to spend any extra money. You’re just getting better results.

How do you reduce churn? You dig in and find out why people are leaving. Are you missing key functionality? Is there a new competitor in the market? Are you too expensive? Is your support terrible? Chances are it’s going to be a combination of things. Interview customers who have recently left, and examine the data for signs that accounts were going to leave before they went.

Finding customers who left your service but are willing to talk to you about why is one of the most challenging problems you’ll face. Offering gift cards or other tokens of appreciation can help, but mostly they’ve moved on. They have very little motivation to help improve a service they no longer use. However, when you can have them, these conversations will be priceless.

You can also try providing exit interviews when people are canceling your service. You can offer a refund on their last payment or other compensation to encourage them to take the time and share feedback. In some cases, they’ll be leaving for reasons outside of your control. Other times, they’ll give you key insights and help you prioritize where to spend your time to reduce churn. This won’t always work, though.

The alternative to reducing churn is good old-fashioned acquiring more customers. If you already have traffic, one of the best ways to stimulate growth is to improve your conversion rates. Let’s say you get ten new customers for every hundred people who sign up: what if you could get five more? You could increase your growth by 50% just by improving and streamlining the initial experience for new customers. This requires having deep insight into your data, but the best way to increase your conversion is to watch people fumble through getting started with your product.

No matter how great your product is, if you haven’t invested serious time in understanding how customers get started with your product, there are hidden problems. Fixing those problems and the resulting experience can work wonders.

Improving Sifter’s initial experience was one of the single most valuable time investments I ever made. Over time, as Sifter changed, I hadn’t kept pace improving it as features were added and others evolved, and it was doing a terrible job. I eventually sat down to spend a week or two designing and implementing some updates, and it got us through one of our plateau phases by guiding more new customers though that first experience.

Once you’ve reduced churn and improved onboarding, you’re ready to focus on increasing the amount of people who show up to your site. Once you reach this point, take some time to read Traction by Gabriel Weinberg, the founder of DuckDuckGo. He details nineteen different tactics and goes into examples of some of the creative ways well-known companies broke through their own plateaus.

One of the trickiest things about growth is recognizing diminishing returns. Cutting churn in half is great, but reducing churn by a minuscule amount is likely a waste of time. The same goes for increasing conversion. Doubling it is great, but at fractional points, it’s going to be increasingly difficult to see further meaningful impact.

If I were starting a new web application today, there are a few things I’d design into the process. First, I’d take the experience for new customers incredibly seriously. I’d spend time working with people and watch them go through the process of setting up their account.

Second, I’d strive to find a key for collaboration and sharing. Of course, this is easier said than done. For instance, if you sign up for a team chat application, it’s almost guaranteed you’re going to invite others. That’s great because it spreads awareness of your product. However, if the value is in collaboration, inviting those new users isn’t enough. You need them to use the application too. Designing this kind of virality into your application can play a big role in growth and reduce the chance of hitting early plateaus. You’ll have to get that virality started, of course, but once it gets going it will invariably help.