Notes on "How I'd Grow Buttondown
A friend sent me Andrea Bosoni's How I'd grow Buttondown, a great and thoughtful piece. It's fun to read someone else do the exact intellectual exercise you've been doing yourself for years, and seeing it from the outside was genuinely useful.
Props to Andrea: the advice is, on the whole, well-considered and reasonable, and I wanted to offer some responses — less to rebut the points and more to characterize them in a more interesting light.
Lean into the demo splash
This is a good idea and one that's been increasingly en vogue. Cursor's landing page is a great example of this pattern done well. And we do plaster links to our demo site throughout blog posts and documentation for exactly this reason.
That being said! An initial A/B test of running our own landing page in this manner showed a pretty bad conversion rate. This was an older, less polished version of our demo site — which has grown substantially more robust in terms of fixture data since then — so it's worth rerunning. But it's a good example of how sometimes intuition gets punched in the face by data. You might be curious about why it didn't perform as well. I am too. My suspicion is that it simply couldn't compete with our prose-heavy positioning copy, which we've iterated and honed pretty sharply over the past few years.
"Emails" is a tough word to sell
The author correctly identifies this, and to a certain extent it's a broader existential crisis we have. If you say newsletters, it's a very specific word that works for some of our customer cohorts but turns off our business users. If you say campaigns, like Mailchimp does, you alienate the prosumer and casual customer cohort — which is our largest by overall volume. In reality, I don't think a lot of people see that H1 tag and are confused unless they're coming to Buttondown from some completely unrelated context, like Hacker News. (See those comments.) Nouns are tricky, though.
Compete pages
Like a lot of programmatic SEO work, these can kind of feel like the simulacra of smart marketing as opposed to the real thing. Our compete pages drive, at this point, very little volume compared to our truer pillar pages — pricing and features. We should invest more in them, sure, but the reality is that they're already richer than 75% of the world's compete pages, and despite that, they drive very little intentional traffic. The juice-to-squeeze ratio is not what you'd hope.
Domain versus subdomain
This is true for a number of reasons, both technical and otherwise. But if we were to ever do a switchover, we wouldn't do it for SEO or marketing reasons — let alone as an experiment — because it's a very serious engineering undertaking, and it's hard to justify very serious engineering undertakings on a hunch that it might have a knock-on effect on search rankings.
Zooming out
All of these are fairly minor points, though. Here's the big one, as it pertains to Buttondown's growth and how I think about marketing writ large.
From both onboarding surveys and overall quantitative traffic: half of our customers come through word of mouth, one-fourth through search, and one-fourth through LLM-based traffic. (I'll talk more about that last bucket in a later post.) When you divide that not by overall customer count but by revenue, it's two-thirds word of mouth and one-third everything else.
This is what used to be referred to in rhapsodizing tones as product-led growth. And it's a lovely little coping mechanism that lets us justify spending the vast majority of our time making a better product and thereby attracting more users. Where this approach falters — and this is the thing I think about most — is trying to expand out of our current orbit of customers to new verticals and cohorts.
Which is to say: most of the advice in Andrea's post is correct and reasonable, and we should do more of it. The reason we don't is not ignorance but prioritization — the ROI on product improvement, for Buttondown's current stage, dwarfs the ROI on marketing optimization. That calculus may change, but to a certain extent we don't want it to — the unit economics and incentives are very clean this way.