For SaaS founders whose growth has slowed or stalled: a data-driven talk that shows exactly what’s capping you and what to actually do about it.
Hitting a ceiling isn’t failure, it’s normal. Facebook’s growth has been largely linear for almost 20 years. With the right mindset, tactics, and structure, founders can turn asymptotes into launchpads for their next phase of growth.
Jason shows why your growth slows and share proven approaches to consider when growth tapers including:
- A conscious shift in focus from explore to execute
- Ruthless Prioritization
- Multi-channel risk mitigation
- Customer concentration control
- Expand into adjacencies
Jason celebrates with you as you shift from hero-driven scrappy beginnings to a disciplined, process-oriented scale machine.
Slides
Find out more about BoS
Get details about our next conference, subscribe to our newsletter, and watch more of the great BoS Talks you hear so much about.
Transcript
Does “if you’re not growing, you’re dying” hold up?
“If you’re not growing, you’re dying.” We’ve been told this. It sounds true. Maybe it is true, maybe it’s just something that sounds smart and isn’t. I’ll come back to this, but for the next 45 minutes let’s assume it is true, and that you’re interested in growing, and that growth is hard sometimes.
This is Lenny Rachitsky’s newsletter. He has a lot of subscribers. He posted this chart on Twitter, a massive spike, and people said, “Holy moly, what happened there?” And he said, “I don’t know. That often happens.” There are probably some people in this room for whom that has happened.
But sometimes growth isn’t mysterious at all. Here’s the growth curve of a company called Projection Lab. People asked what happened at the inflection point, and the founder said: “John started.” Who’s John? John’s a marketer. It turns out that if you make something and don’t tell anyone, it doesn’t sell that well. And if you make something and do tell people, it might. Not that mysterious.
One thing is certain, though: growth always slows down. Always. It’s a kind of law of physics, which I’ll explain. And then I’ll explain what to do about it.
Why customers leaving feels personal
The first thing I want to talk about in terms of growth is when people leave. Because I feel this personally. Forget the finance side for a moment. Think about what had to happen for a customer to even be there. Someone had to see something you put online. What percentage of people click through a link? Point one percent? It’s almost impossible. Then they didn’t bounce off the homepage in three seconds. They thought the features were reasonable. They weren’t scared by the pricing. They bought it. They actually onboarded. They were there for a year. After all of that, they left.
That is like telling me my product is terrible. They got through that entire gauntlet. What the hell?
There’s also a saying I use: cancellation wins. Not only is it personal, it controls your growth rate. Let me explain that.
The math: marketing is linear, cancellations are exponential
Here’s new customer growth at ConvertKit (now called Kit) over roughly a ten-year period. You can see something like a line, more and more people added each month, with wobbles, but basically linear. And here’s why.
When you figure out a marketing channel, you figure it out, you get however much you can get out of it, you spend a period optimising, and then everyone’s already seen it. They’ve searched the thing. If they were going to click the ad, they probably already would have. Then often the channel declines, because people have seen the ad seven times, they’re tired of you, or competition in the auction heats up. AdWords goes to the highest bidder, which is often the least rational one. Or the channel declines altogether, the way magazine advertising has. When you stack more marketing campaigns on top of each other, you get more volume, but still the same shape: linear.
Cancellations don’t work like that. Completely different.
Suppose you have a company with 1,000 customers and marketing is adding 100 per month. Cancellation rate is 5%, so 50 customers leave each month. Net, you’re growing by 50 a month. Fine.
Now imagine that same company triples overnight. Marketing still adds 100 per month. AdWords doesn’t care how big you are, the channel doesn’t care. But cancellation does care. 5% per month means 5% of your size, and your size has tripled. Now this company is shrinking. Marketing didn’t change. Size alone makes cancellations grow, definitionally.
Marketing grows linearly. Cancellations grow exponentially. That’s what a percentage is: it’s exponential. The bad things being exponential is not good.
The growth ceiling, and how to measure it
This creates what I call a growth ceiling. Every company has one. And you can calculate yours exactly.
I call this metric Max MRR: the maximum MRR you could reach before cancellations catch up with new customer adds. The formula is simple. We want to know when cancellation MRR equals new MRR. Cancellation MRR is just size multiplied by monthly cancellation rate. So:
Max MRR = New MRR added per month / Monthly cancellation rate
If you’re adding £150k in new MRR per month and your monthly cancellation rate is 5%, your ceiling is £3 million. That’s it. That’s as big as you can get.
This is part of what happened at Buffer. Buffer’s monthly cancellation rate was 5%, not hypothetically, actually. And so cancellations caught up. Of course there’s more to the story, and Joel will tell it far better than I can. But the math is the math.
This did not happen at ConvertKit. Here’s their actual MRR over ten years: roughly linear, with bumps, but it kept going. How? Because they kept intentionally pushing their cancellation rate down. It’s now around 3%. They didn’t stumble into it. They worked on it continuously, on purpose.
Max MRR as a predictive tool
Here’s why I think tracking Max MRR is worth your time. You can plot it alongside your MRR every month. When there’s a blip in your revenue curve, it can be easy to explain away in a board meeting. But Max MRR reacts much faster and much more visibly, because it’s directly tied to the underlying drivers.
At ConvertKit, there were two separate periods where revenue dipped. In one, cancellations spiked. In another, new MRR dropped. In both cases, the revenue dip was subtle, easy to explain away. But Max MRR dropped sharply and obviously. You can’t get away from it.
At Buffer the picture is even clearer. For a long time things were going well. But if you’d been plotting Max MRR, you would have seen the problem well before the revenue line gave you any signal. And at the other end, when things get fixed, the Max MRR line pops up months before the revenue line confirms it’s working. You’d know within a month or two that the fix is real, rather than waiting six or twelve months for the revenue to confirm it.
A lot of people say cancellation rate is hard to internalise: it’s just a number, hard to feel. Max MRR is visceral. This is the ceiling. This is where we’re heading. I think it’s a better way to communicate the same underlying reality.
Want more of these insightful talks?
At BoS we run events and publish highly-valued content for anyone building, running, or scaling a SaaS or software business.
Sign up for a weekly dose of latest actionable and useful content.
Unsubscribe any time. We will never sell your email address. It is yours.
Why cancellation feedback misleads you
The obvious response is: find out why customers are cancelling, and fix it. So you build a dropdown on the cancellation screen with options like “too expensive,” “missing feature,” “found a better solution.” People pick things, and you see what’s most common.
At WP Engine, “too expensive” was by far the most common response. Then I noticed it was also the first item in the dropdown. I randomised the order, and all of a sudden every option was picked roughly evenly. They weren’t telling me anything. It was useless. Of course, by the time they’re cancelling, they’ve already mentally checked out. Why would they spend any time helping you? They’re done.
A company called Groove found a better way. Same problem, same dropdown, same useless data. They started sending open-ended emails instead. Most variations didn’t work. Then they hit on one.
The subject line: “quick question”, not capitalised, informal. Signed from Adam, the founder, no job title listed. Three sentences. No corporate footer. Just: why did you cancel? Their usable response rate, meaning actual actionable feedback, was 10%. That’s genuinely good for this kind of thing.
Then they tried changing the wording slightly: “What made you cancel?” instead of “Why did you cancel?” One word, asking for a thing rather than a reason. That doubled the usable response rate to 20%. Now you’re getting somewhere.
The deeper cause is never what they say
Even with better feedback, there’s a structural problem. Think about “too expensive” again. That customer saw your pricing, decided it was fine, and bought the product. Then they cancel saying it was too expensive. That’s a contradiction. Why did you buy it at all?
What actually happened is something like this: they wanted a specific feature, or expected the product to do something it didn’t do, or couldn’t get it integrated with their tool, or couldn’t get it working well enough to deliver value. And because of that, it felt like it wasn’t worth what they paid. So they said “too expensive”. That’s not the cause. That’s the symptom.
In medicine, when someone dies, there’s a form you fill in asking for the proximate cause of death: the last thing that happened. Stopped breathing. True. But if you ask why, you get: they were in an accident, their injuries were too severe. Why were they in an accident? They passed out at the wheel. Why? Undiagnosed diabetes. The useful, actionable reason is three levels down.
Cancellation feedback works the same way. If you respond to “too expensive” by lowering prices or moving features to a lower tier, you’ve treated the symptom. The customer already told you price wasn’t the issue: they bought the product. The real problem is somewhere deeper, and you’d never find it by taking the feedback at face value.
Find them before they cancel
If customers in the process of cancelling are unreliable informants, the solution is to find them before they’ve made that decision.
At WP Engine we talked about the happy path. It’s not an original phrase, but it’s a useful one. The happy path is the sequence of actions a customer takes when things are going well: they signed up, they added their site, they deployed, they’re getting results. Whenever a customer falls off that path, perhaps they haven’t done the next expected thing, or they’re stuck, or they’ve gone quiet: that’s a signal. They haven’t cancelled yet. You can still have a conversation.
Don’t wait for the cancellation. Cancellation is the worst time to have that conversation. Even with a well-crafted email, you’re talking to someone who has already decided. The happy path lets you intervene while there’s still something to save.
Onboarding: where you get the biggest return
I made a video a few years ago called the Profit Whale. YouTube shows you exactly what percentage of viewers are still watching at each second. Mine showed the typical shape: 30 seconds in, half the audience has gone. Not great for the ego.
But here’s the thing. If I improve the opening, even a little, because the drop-off there is so steep, I could double the number of people who watch the whole thing. Improving the back end of the video barely moves the needle, because so few people are there by that point.
Customer cohorts in software work the same way. If your onboarding improves even modestly, the number of customers who stick around for five years could double. The lifetime value impact of fixing onboarding is almost always larger than anything you can fix later in the customer journey.
Onboarding is almost always where you get the most value per unit of effort on cancellation.
Getting your existing customers to grow you
We’ve been talking about cancellations, one side of the growth ceiling. What about the other side? Can we make marketing grow proportional to our size, rather than just proportional to our spend?
The answer is yes. There are sources of customers that scale with your customer base.
Give-and-get referrals
The standard referral programme, invite a friend and get £75, I dislike this. It turns your customer into a weird salesperson. Imagine them calling a friend: “Hey, I’m going to get money if you sign up.” That’s not a natural conversation.
The better version is what I call give-and-get: both the giver and the getter receive something of value. Dropbox did this well. Invite a friend and you both get 500 megabytes of extra storage. Now the customer calls a friend and says: “If you sign up for this thing I use, we both get more space for free. We’re both getting a good deal.” Completely different conversation. The founder of Dropbox has said many times that this programme was the reason for their explosive early growth.
Asking for advocacy at the right moment
You can simply ask customers to leave a review or say something online. Most companies never do. At WP Engine, when we started asking customers to rate us on TrustRadius, our score improved significantly, just because we asked.
But when you ask matters. Don’t ask at the very beginning, before you’ve delivered value. They’re probably leaving at that point. Don’t ask at the very end of the customer lifecycle, when the fewest people are left.
Ask at a moment of genuine success: the first time they get a real result from the product; when they hit a meaningful milestone (the 100th post scheduled in Buffer, the 100th attack blocked, whatever fits your product); right after a strong support experience; when they join a beta programme; when they get an award for something your product helped them build. Dropbox times their referral prompt precisely: when you’re running out of storage, they surface it and say you can get more space without paying by referring a friend. That’s when the pain is real, and so the motivation is real.
Net Revenue Retention: the only way to build a large company
There’s one more way the remaining customers can offset cancellations: they can pay you more over time. This is Net Revenue Retention.
NRR tracks a cohort of customers over 12 months. At the start, they represent a certain MRR. At the end, some have upgraded, some have downgraded, some have cancelled. Whatever the net result is, that’s your NRR. Start with £100k, end with £75k: 75% NRR. Start with £100k, end with £110k because upgrades outpaced everything else: 110% NRR.
The difference in what this means for growth is dramatic. At 75% NRR, even with consistent new customer adds, you hit a ceiling fast. At 100% NRR, cancellations and upgrades exactly balance out, and you grow linearly with your marketing. Above 100% NRR, the existing customer base grows on its own even without new customers. I’ll state this plainly: that last scenario is the only way to build a very large company. Every SaaS company that has gone public had positive NRR. The median at IPO is around 119%.
Dropbox is a vivid example. They’ve been acquiring fewer new customers every year for years. But their NRR means their total revenue keeps growing, all the way to $2 billion. Marketing is declining. The cohorts are doing the work.
To get NRR above 100%, customers need to grow with you. Two-axis pricing helps: pay more for more features, and pay more as you use more. Slack does both. But the less obvious piece is showing customers the value you’re delivering, so that price increases feel earned rather than arbitrary.
A security product dashboard that shows how many attacks were blocked this month, and that it’s 5% more than last month, primes the customer to feel that the product is doing more and more for them. When they eventually hit a pricing threshold, it doesn’t come as a surprise. If you’ve never shown them what you’re doing, you’ve become invisible. The value has to be visible, not just real.
When to expand: the adjacency matrix
At some point, you may have extracted as much as you can from the current product and market. That’s when to think seriously about expansion: a new product, a new customer segment, a new geography. This is a significant decision and I want to share the framework we used to think about it.
At WP Engine, we started noticing enterprise-scale companies buying our product: one of the biggest banks in the world, one of the biggest software companies. We wondered whether there was a real opportunity in enterprise. Two ideas came up.
One: targeted marketing campaigns for enterprise buyers, landing pages and enterprise messaging, without changing anything else. Same product, same price, same engineering. Low investment, something to learn from.
Two: build a full enterprise product. New pricing, dedicated engineering investment, added governance and compliance features, a whole division devoted to it.
How do you choose? I use what I call the adjacency matrix. List the functional areas of your company: marketing, sales, engineering, business model, unit economics, support. For each proposed initiative, rate how much change each area requires. Three options only: trivial (business as usual, they’ll barely notice), project (within our wheelhouse, we’ve done this kind of thing before), or overhaul (we don’t have the expertise, we’ll need to hire or retrain, it’ll probably take twice as long and cost twice as much as we think).
The matrix doesn’t make the decision for you. It does two things. First, it forces each functional area to say honestly what the change means for them. People from other departments almost never know this without being asked directly. Second, it surfaces the question of what you can actually execute right now, given the real state of each team.
If your sales team is in trouble, this is not the moment to take on a new initiative that requires an overhaul of sales. But if your engineering team is healthy and energised, an engineering-heavy initiative might be exactly right. The matrix tells you which of your options aligns with where you’re strong and avoids where you’re weak.
It also gives you a clear way to explain the decision to the company. “We’re not going into Japan right now because our marketing and finance teams aren’t in a position to handle it” is a real explanation. People understand it. It aligns the team in a way that a scoring spreadsheet never will.
So: if you’re not growing, are you dying?
Let me come back to where I started.
My instinct is always to be contrarian: to say that’s a bumper sticker, probably something a VC says to make founders work harder. But I actually think it’s true, not just as a business statement but as a felt experience.
When you’re in a situation of real decline: morale is low, competitors are winning, you might be letting people go, everyone wonders who’s next, there’s blame and argument and uncertainty. And you as a founder didn’t sign up for this. You signed up to build something, to create something, to have something that works. That’s a bad situation for everyone involved, and it’s bad for reasons that go beyond the numbers.
So I think growth matters. But growth doesn’t have to mean revenue. If revenue growth is hard, profit is much more within your control, and a profitable company has options: money to act on, employees who aren’t worried about their jobs, a founder who can actually afford to be in it for the long term. Maybe the statement isn’t “grow revenue or die” but “grow profit or die.”
There are mission-based companies where growth means impact. Customers of those companies tend to be deeply loyal. Employees often choose them over higher-paying alternatives because the work means something. If the mission is growing, that’s something real.
And there’s the personal dimension. If you as a person are not growing, not learning, not being stretched. For the kind of driven person who starts a company, that may be its own kind of problem.
So I’ll leave you with a question, whether you direct it at your company or at yourself: what are you going to do next to grow?
Want more of these insightful talks?
At BoS we run events and publish highly-valued content for anyone building, running, or scaling a SaaS or software business.
Sign up for a weekly dose of latest actionable and useful content.
Unsubscribe any time. We will never sell your email address. It is yours.
Q&A
Mark Littlewood: We’ve got some time for questions. Carl, go ahead.
Carl Ryder: Great talk, as always. Two questions. First: you talked about cancellation surveys not working and combo boxes being useless. Has anyone tried giving customers the option to gift a feature to a friend, so in the moment of cancelling they reveal which features they actually value by choosing what to give away. Second: a lot of people are sceptical of paid marketing because they worry it acquires customers who churn. You’ve done a lot of work on knowing who your best customer actually is. Could you talk about that?
Jason Cohen: On the gifting idea: I love it. The idea of revealing people’s true preferences through a concrete choice like that is genuinely interesting. One practical constraint: for a customer to gift a feature, it probably has to be something they have that others don’t, which limits it to customers on higher tiers. But the principle of surfacing what people actually value through a real decision rather than a stated preference is valuable. You could follow up with the people who made choices and ask why. Even a 10% response rate would give you something useful.
On paid marketing and churn: I find all marketing channels can produce poor customers if you target badly. If you spray broadly, you get mixed results. If you’re precise about who your best customer is, paid marketing can produce excellent customers. The question is whether you know who that is. There are people here who are genuine experts on that question, and I’m happy to talk about it separately.
Mark Stephens: You mentioned adding features customers will pay for, and competitors changing the landscape. How do you think about competitive response: features they ship, market expectations they shift?
Jason Cohen: I wouldn’t ignore competitors, but I wouldn’t make decisions based on what they do either. Half the time, when you get under the covers of a competitor’s business, the things that look brilliant from the outside are actually losing them money. Chasing them is a bad idea.
What matters is what customers want. If a competitor ships a feature and your customers start mentioning it, and some start leaving because of it, you pay attention. But you pay attention because the customer told you, not because the competitor did. The competitor may have changed the expectation. That’s fine. You’re still following the customer.
The analogy I use is football: don’t watch the feet, watch the hips and body. You follow the commitment signal. Your customers are already looking at your competitors, reading reviews, comparing options. Since they’re looking, you have to look too. You can’t understand your customers if you’re not seeing what they’re seeing. But the question always comes back to: what does the customer actually want?
On AI specifically: nobody really wants AI in their product in the abstract. What they want is more of what they always wanted: more leads, better closed sales, faster output. If AI delivers more of that, great. But “add AI because the competitor added AI” is not a reason.
Audience Member: You touched on profit at the end. Have you seen situations where better customer service drives cancellation down and revenue growth up, especially in commoditised markets?
Jason Cohen: Can great customer service reduce cancellation and cause customers to pay more? Yes, but there’s a distinction that matters.
If your customer service is good because the product isn’t, with customers calling constantly because things don’t work or they can’t figure things out, that’s a band-aid, not a strategy. It will temporarily slow cancellations, but ultimately customers leave because the product isn’t what they need, and no amount of good service changes that.
But if the product is solid and you’re adding consultative service on top, helping customers think through their actual problem and pointing them toward solutions they hadn’t considered: that’s genuine value. At WP Engine, customers would ask: “Who should I hire to do X?” or “Which plugin should I use for Y?” That’s not covering for product problems. That’s expanding what the product means to the customer.
There’s also interesting data on this. NPS is not correlated with loyalty, but not needing to contact support is correlated with loyalty. The customers who stay are often the ones who rarely have to call. That suggests the goal of customer service isn’t to make people happy after something goes wrong. It’s to build a product where things rarely go wrong in the first place. If you have that, and you layer great service on top, that’s a real differentiator. But fix the product first.
Mark Littlewood: Final question. Jose, go ahead.
Audience Member (Jose): What are your favourite sources of education? Favourite book?
Jason Cohen: On what topic?
Audience Member (Jose): Finding product-market fit, I suppose.
Jason Cohen: The outline of what you need is knowable. A lucrative market, people who know they have the problem, real urgency, some inciting event that makes them act now rather than later. Bob Moesta has talked about this at previous BoS conferences, and those talks are worth watching. The product has to solve the problem well, your homepage has to make the right customer immediately recognise it’s for them, and your marketing has to reach them at the moment of that inciting event.
You know all of that. And yet most people don’t find product-market fit. And when you do, you often can’t fully explain why it clicked when it did.
My broader view on advice: for almost every recommendation, someone credible says the opposite. I can find companies that succeeded doing X, and other companies that succeeded doing the opposite. Both the four-by-four squares are full, and you can’t use success stories to decide.
So what do you do? Find an approach whose reasoning actually resonates with you, not just the conclusion, but the way the thinking works. That gives you something you can execute well, because it fits how your mind works. Stay consistent within it. Consistency in a coherent approach beats jumping between frameworks every quarter. And look for the version of the advice that makes you a better version of who you already are, not one that asks you to become a different person. If that’s the feeling when you read it, that’s a signal worth following.
Jason Cohen
Jason has built four software startups, both bootstrapped and funded, both alone and with co-founders. All of them grew to more than $1m annual revenue.He sold two, and currently serves as CTO of the fourth, WPEngine, with 380 employees headquartered in Austin, Texas. More recently, he has also been an angel investor and was a founding member of Capital Factory, an Austin incubator and co-working space. He writes about software and startups at ASmartBear.
Next up
Online workshops
Can’t make it? More dates coming up