LLMs and Buttondown
Buttondown has grown a lot in 2026; some of that growth is uncomfortable. I've talked to enough people about this in real life that it felt worth writing some of it down while it's fresh — if for no other reason than to revisit in a few years' time.
First, the headline. Our month-over-month growth rate in Q1 2026 was double our growth rate in Q4 2025. Buttondown has, roughly, grown a little less than 2x every year of its existence; this — its eighth year — is poised to shatter that, if trends hold.
Almost all of that incremental growth, meaning the growth in addition to our historical trend, I attribute to LLMs. We ask people when they sign up what brought them here, and an answer that went from surprising to banal to overwhelming over the course of Q1 was: an LLM. Users of all stripes cite an LLM as the reason that they ended up at Buttondown's front door.
These users, broken into their own cohort, have some interesting numerical properties. They both activate (i.e. send their first email) and convert (i.e. start paying us) at a worse but not dramatically worse rate than the median. Here are some numbers: One thing worth calling out: while the LLM-born cohort skews more B2B than our historical B2C-leaning median, it is remarkable how few of them are evocative of the stereotype. Teachers, public servants, and nonprofits are all extremely well-represented. Make of that what you will.
| Jan | Feb | Mar | |
|---|---|---|---|
| % of new users citing an LLM | 15% | 32% | 49% |
| Conversion rate, relative to median cohort | 85% | 71% | 75% |
It feels weird to complain about this. So let me list some of the weird ways it has surfaced.
1. Churn. It is too early to really understand the statistical significance of this, but I'm faintly worried about churn being higher for these users than for our traditional cohort. A lot of our approach to onboarding and pricing hinges on a fairly high LTV; it's how we can justify outsized investment in support, for instance. Definitionally, lower LTV warps a lot of these decisions.
2. Support. While the absolute volume of support tickets coming from LLM-born users isn't significantly higher than the median, the shape of those tickets is off. To put it bluntly: a lot of the tickets we get are themselves LLM-generated. This is, frankly, extremely annoying — and demoralizing for me and the team to spend half an hour meticulously answering some complex question only to receive a machine-generated reply in return.
3. ICP. Buttondown has grown through word of mouth alone since its inception, which means that our customers all kind of have the same vibe — and, more importantly, know what our vibe is in return. LLM-born users, by virtue of having outsourced the research process, really don't. This materializes in every aspect of the conversion pipeline, from prospecting (I recently had a sales call with a user who was shocked to learn we don't support cold email) to volume.
People have asked why I think we have been the beneficiary of this genre of growth. There is one fairly interesting reason: we have accidentally built a very LLM-friendly business in this space. While we don't have an MCP, and our CLI is fairly nascent, an early design decision — inspired by my time at Stripe — was to dogfood the entire API in the core dashboard. As a result, our REST API is extremely well-documented and feature-rich, because it has had the constraint of needing to power the core user experience. This is a flip from most other competitors, where the API is vestigial at best.
I mean this in a positive light, mostly.
I led this post by saying I feel uncomfortable about the growth, and it's true. The business is growing faster than ever before. It is also being taxed in a way that it has never really been taxed. More than that: it is the first time I really feel like we've had a loss of agency around driving the business — a sense that, well, sometimes you just run the same direction as everyone else.
Of course, we don't really plan on doing that. What we plan on doing is roughly as follows:
- Pull forward future investments in internal tooling and infrastructure, to proactively scale both our processes and our systems.
- Invest even further in making the API really, really good. We have already started to see just how far API- or LLM-driven usage can go. For instance, we've had a number of bugs I am a sicko who thinks that bugs being revealed is a good thing. because parts of the API I once considered esoteric were not handling JSON schema correctly — we were relying naively on the front end of the core app to do that work for us. We're investing in Schemathesis and other automated fuzzers to firm this up.
- Firm up some of our support policies, especially around our interactions with prospects and free users, in order to protect our support engineers' time and energy.
- To paraphrase my friend Nick: treat the new revenue like it doesn't exist until enough time has passed to be absolutely sure about things.
Odd times ahead; odd times right this very second.