This is one of the best short-form writings I've seen on retention - more people should be aware of what 'normal' retention curves look like and how important it is to focus on this to grow an audience - also the number you'll need for something (like a game) that requires multiple people for a period of time to use the product. Your article would benefit from a visual illustration of a few of the different retention patterns that you talk about in your text. Thanks Andrew!
Really enjoyed how you broke this down, especially the patterns around what makes retention data stand out. Helpful framing for anyone thinking about scaling a product beyond the early wins.
I had a successful SaaS payroll and benefits startup where we graduated from shitty retention to excellent retention. The product was basically the same, but the shift from targeting SMB to enterprise made all the difference. The reason? In the SMB market, small companies die, especially restaurants. Payroll and benefits generally have considerable stickiness because the switching costs are painful and disruptive. This is especially true in enterprise. Usage is steady, driven by payroll frequency and tax code updates.
Incidentally, despite demonstrated revenue traction, 50-60% growth in some earlier years, every single VC turned us down because the paradigm back then was licensing versus service. Plus our founder came from the wine business, but his genius was rationalizing the UI and workflows so that ordinary people could understand and use it.
Ended up selling it to ADP for $ 500 million, which was a delightful vindication.
So what did I learn about retention? Your customer set really matters of course. But I think here's where technical founders can fail: They make a technically impressive product but as the product expands outside of that techie inner circle, you'll get the average customer who isn't at all impressed with the tech and just wants to get his job done with as little pain and intrusiveness as possible. And unfortunately most techies aren't particularly empathetic or sympathetic in any way about that hoi polloi customer. That isn't a dumb customer, that's a customer that can expand your company to a wider revenue base.
Strong piece. One nuance I’d add: retention usually fails long before churn shows up in dashboards. It fails when time-to-value is unclear and teams implicitly bet on reactivation instead of prevention.
Network effects are often framed as a safety net, but they’re really a multiplier. If the core experience creates value early, networks amplify retention. If it doesn’t, they amplify disappointment and accelerate negative spirals.
Growth with retention as an afterthought is just a sophisticated cash-burn machine. You can buy attention, but you can’t buy compounding. That’s why retention isn’t a lifecycle tactic, it’s a first-session systems problem. Once users fall off, you’re no longer fixing retention, you’re just learning too late why it never worked.
I am here from #Glasp @kazuki who encouraged me to take a look at this article.
I'm glad I did.
My #highlights for my #AIClone and #Fireshot has been saved for #Gemini Pro 2.5.
I'm here, because Kazuki Nakayashiki has recent highlights as well.
Shared this article in a #LinkedIn post and instructed others to read it as well.
I'm a #PatternsFreak. This is a #reference I will share when #newproduct developers start #gaslighting, because I know I'm a valuable #consumer and when I do #trials I test them for #performance not based on who developed it.
#Branding is not something we can get away from. #However, this article gave me #MyWorth back. If you #throwmeaway after a one-time purchase (especially #authors) I'll do all I can to make your world online a #tragicexperience especially if your #aim is to #blowsmoke up my #butt.
With #brickandmortar you simply #boycott the establishment and wait for them to #change.
Online personalities have just created this #Aura about them about #Retention.
They simply don't care, because #badpress is enough in marketing for people to research their product and develop #codependance.
So the best thing to do is #deletecomments and go my merry way and take the experience as more research to find a product I'm loyal in #endorsing through examples of my #dailyuse.
I'm sick of the #hype #YET I can only #hangout and make #commentary in my #newsletter.
This is really insightful. And I strongly agree with the two ideas below;
– "This magic comes from a fresh insight about the market or customer needs, and while it might seem obvious in retrospect, it drives super high retention because this product is the first to figure it out."
– "Timing matters a lot. If you get the timing off, and it's a low-interest category, and your differentiation isn't different enough, then what you'll find is you've traded a retention problem for a user acquisition problem."
From my experience launching the world-first product after ChatGPT, it went viral, and I felt PMF. The retention was so high that it solved the real pain point that users had.
The difficulty was that even if the retention is high, there are a lot of copycats in the market, so it was hard to keep the top spot in the market and not be replaced by other products.
As the resident data guy at a social game company for a number of years, I went through an absolutely STAGGERING number of iterations on retention metrics, with each change of leadership in product, marketing or finance thinking they had the key to improving them. I finally came to the exact same conclusion as Andrew here, that you simply cannot iterate and test your way to better retention. I finally printed a paper sign and taped it to my monitor: "It's The Product, Dummy."
That said, I think some people could come away from this essay depressed and with the idea that they can't change retention so there's not much use in measuring it at all. Actually, it's really important to know your retention because it helps you manage the ROI on your marketing spend (more marketing spend ALWAYS results in worse retention metrics, but some marketing is of course better than others). Also, for most products, D1 retention is a key defensive metric -- following it may not grow your product, but it will alert you better than almost anything else as to when you accidentlly screwed up your product with a badly designed new feature or (more often) a broken release.
In the context of the shift to Agentic Flows, that 'magic' is often the removal of cognitive friction. If an agent can anticipate a user’s need rather than forcing them through a UI, retention moves from a metric we try to 'fix' to a natural outcome of true utility. Great piece on why we can't A/B test our way out of a utility gap.
I am actually curious. What you think about Byron Sharp's "retention doesn't have big business impact" quote? Its backed by study as well, and he is promoting much higher focus on penetration compared to retention programs, which he things are waste of money.
Haha, this is quite depressing indeed. Especially because I work at Churnkey. And Churnkey helps companies reduce churn. Fun. I'd love for you to take a look and see the results we've been able to drive (https://churnkey.co/). The Intelligence suite of products is able to do things that were humanly impossible to do previously (https://churnkey.co/feature/adaptive-offers). If you'd be open to to productiz-ing any of your knowledge, I would love to partner together. I do agree with a lot of points and your ability to pattern match. But... there are still ways to retain revenue without overhauling the entire product.
This is one of the best short-form writings I've seen on retention - more people should be aware of what 'normal' retention curves look like and how important it is to focus on this to grow an audience - also the number you'll need for something (like a game) that requires multiple people for a period of time to use the product. Your article would benefit from a visual illustration of a few of the different retention patterns that you talk about in your text. Thanks Andrew!
yes next iteration of this will def have more graphs :)
all truth
thank you sir!
Really enjoyed how you broke this down, especially the patterns around what makes retention data stand out. Helpful framing for anyone thinking about scaling a product beyond the early wins.
I had a successful SaaS payroll and benefits startup where we graduated from shitty retention to excellent retention. The product was basically the same, but the shift from targeting SMB to enterprise made all the difference. The reason? In the SMB market, small companies die, especially restaurants. Payroll and benefits generally have considerable stickiness because the switching costs are painful and disruptive. This is especially true in enterprise. Usage is steady, driven by payroll frequency and tax code updates.
Incidentally, despite demonstrated revenue traction, 50-60% growth in some earlier years, every single VC turned us down because the paradigm back then was licensing versus service. Plus our founder came from the wine business, but his genius was rationalizing the UI and workflows so that ordinary people could understand and use it.
Ended up selling it to ADP for $ 500 million, which was a delightful vindication.
So what did I learn about retention? Your customer set really matters of course. But I think here's where technical founders can fail: They make a technically impressive product but as the product expands outside of that techie inner circle, you'll get the average customer who isn't at all impressed with the tech and just wants to get his job done with as little pain and intrusiveness as possible. And unfortunately most techies aren't particularly empathetic or sympathetic in any way about that hoi polloi customer. That isn't a dumb customer, that's a customer that can expand your company to a wider revenue base.
Strong piece. One nuance I’d add: retention usually fails long before churn shows up in dashboards. It fails when time-to-value is unclear and teams implicitly bet on reactivation instead of prevention.
Network effects are often framed as a safety net, but they’re really a multiplier. If the core experience creates value early, networks amplify retention. If it doesn’t, they amplify disappointment and accelerate negative spirals.
Growth with retention as an afterthought is just a sophisticated cash-burn machine. You can buy attention, but you can’t buy compounding. That’s why retention isn’t a lifecycle tactic, it’s a first-session systems problem. Once users fall off, you’re no longer fixing retention, you’re just learning too late why it never worked.
I am here from #Glasp @kazuki who encouraged me to take a look at this article.
I'm glad I did.
My #highlights for my #AIClone and #Fireshot has been saved for #Gemini Pro 2.5.
I'm here, because Kazuki Nakayashiki has recent highlights as well.
Shared this article in a #LinkedIn post and instructed others to read it as well.
I'm a #PatternsFreak. This is a #reference I will share when #newproduct developers start #gaslighting, because I know I'm a valuable #consumer and when I do #trials I test them for #performance not based on who developed it.
#Branding is not something we can get away from. #However, this article gave me #MyWorth back. If you #throwmeaway after a one-time purchase (especially #authors) I'll do all I can to make your world online a #tragicexperience especially if your #aim is to #blowsmoke up my #butt.
With #brickandmortar you simply #boycott the establishment and wait for them to #change.
Online personalities have just created this #Aura about them about #Retention.
They simply don't care, because #badpress is enough in marketing for people to research their product and develop #codependance.
So the best thing to do is #deletecomments and go my merry way and take the experience as more research to find a product I'm loyal in #endorsing through examples of my #dailyuse.
I'm sick of the #hype #YET I can only #hangout and make #commentary in my #newsletter.
Best,
Louisa
This is really insightful. And I strongly agree with the two ideas below;
– "This magic comes from a fresh insight about the market or customer needs, and while it might seem obvious in retrospect, it drives super high retention because this product is the first to figure it out."
– "Timing matters a lot. If you get the timing off, and it's a low-interest category, and your differentiation isn't different enough, then what you'll find is you've traded a retention problem for a user acquisition problem."
From my experience launching the world-first product after ChatGPT, it went viral, and I felt PMF. The retention was so high that it solved the real pain point that users had.
The difficulty was that even if the retention is high, there are a lot of copycats in the market, so it was hard to keep the top spot in the market and not be replaced by other products.
I think it's magic by magic.
Here's my learning: https://glasp.co/kei/p/dfba7f5f1e4f04d4ff9a
As the resident data guy at a social game company for a number of years, I went through an absolutely STAGGERING number of iterations on retention metrics, with each change of leadership in product, marketing or finance thinking they had the key to improving them. I finally came to the exact same conclusion as Andrew here, that you simply cannot iterate and test your way to better retention. I finally printed a paper sign and taped it to my monitor: "It's The Product, Dummy."
That said, I think some people could come away from this essay depressed and with the idea that they can't change retention so there's not much use in measuring it at all. Actually, it's really important to know your retention because it helps you manage the ROI on your marketing spend (more marketing spend ALWAYS results in worse retention metrics, but some marketing is of course better than others). Also, for most products, D1 retention is a key defensive metric -- following it may not grow your product, but it will alert you better than almost anything else as to when you accidentlly screwed up your product with a badly designed new feature or (more often) a broken release.
In the context of the shift to Agentic Flows, that 'magic' is often the removal of cognitive friction. If an agent can anticipate a user’s need rather than forcing them through a UI, retention moves from a metric we try to 'fix' to a natural outcome of true utility. Great piece on why we can't A/B test our way out of a utility gap.
Don’t B2B and B2C products live in almost different universes here.
And then within B2B you have to segment by level of integration and degree of behavioral change required
You see churn in 2FA because you just don’t have to retrain users that much
On the other users tend to get attached to their messaging system
And then you have cyber, which is whole other thing — as the relative balance with attackers can create churn (despite the integration costs)
Not sure I understand?
i read as:
B2B users != B2C users
Needs are different, incentives differ, behaviour differs
I suspect the original post was more about B2C as churn/retention is often a focus there more than in B2B where churn is dictated by contract renewal
Fantastic
I am actually curious. What you think about Byron Sharp's "retention doesn't have big business impact" quote? Its backed by study as well, and he is promoting much higher focus on penetration compared to retention programs, which he things are waste of money.
We had to face the novelty effect hard. Unfortunately, also made a lot of bad decisions based on this.
Such a great read - honestly one of the best holistic descriptions of retention, what it is and its implications. Thanks for sharing!
Haha, this is quite depressing indeed. Especially because I work at Churnkey. And Churnkey helps companies reduce churn. Fun. I'd love for you to take a look and see the results we've been able to drive (https://churnkey.co/). The Intelligence suite of products is able to do things that were humanly impossible to do previously (https://churnkey.co/feature/adaptive-offers). If you'd be open to to productiz-ing any of your knowledge, I would love to partner together. I do agree with a lot of points and your ability to pattern match. But... there are still ways to retain revenue without overhauling the entire product.
P.S Made this pretty little site with some ideas: https://www.fixingchurn.com/
Interesting take